{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rschmit, mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"The device reports rschmit, mkhan, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall outbound policy\", \"finding\": \"Firewall outbound policy is no firewall instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Compliant\", \"title\": \"BitLocker is compliant\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Compliant\", \"title\": \"Account lockout duration: meets the baseline\", \"finding\": \"Account lockout duration is set to 30 minutes, in line with the required >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (windows):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"Found: no AI tools found. This satisfies the baseline requirement of inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nHOSTNAME: LAPTOP-K5GG34O\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=15, Non-Compliant=42, NotApplicable=3, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 8\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Non-Compliant | 5 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout configured\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Non-Compliant | service True, antivirus False\n- malware.realtime | Real-time protection | Non-Compliant | off\n- malware.on_access | On-access protection | Non-Compliant | off\n- malware.tamper | Tamper protection | Compliant | on\n- malware.signatures | Security intelligence | Compliant | 0 days old (version 1.433.847.0)\n- firewall.domain.enabled | Domain firewall | Non-Compliant | off\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Non-Compliant | allow\n- firewall.private.enabled | Private firewall | Non-Compliant | off\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Non-Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | block\n- screen_lock.enabled | Screen lock | Non-Compliant | no machine inactivity limit; no signed-in user to check\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | screen saver does not require sign-in for: CONTOSO\\akaya\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | CONTOSO\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | LAPTOP-K5GG34O\\rschmit\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (CONTOSO\\akaya)\n- encryption.feature | BitLocker | Non-Compliant | off\n- encryption.enabled | Encryption | Non-Compliant | EncryptionInProgress (42%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, None\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 1 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 27 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | automatic updates turned off by policy\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 227 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26100\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: success; account management: none\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Non-Compliant | enabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | chromeremotedesktop\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Non-Compliant | Recall feature Enabled, not disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.min_age: Low\n- password.complexity: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.service: High\n- malware.realtime: High\n- malware.on_access: Medium\n- firewall.domain.enabled: High\n- firewall.domain.inbound: High\n- firewall.domain.outbound: Low\n- firewall.private.enabled: High\n- firewall.private.inbound: High\n- firewall.private.outbound: Low\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.feature: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.human_oversight: Medium\nSCORE (computed): compliant=15, assessed=57, percent=26\nTOP GAPS (computed, highest risk first):\n- malware.service | Microsoft Defender Antimalware | High | service True, antivirus False\n- malware.realtime | Real-time protection | High | off\n- firewall.domain.enabled | Domain firewall | High | off\n- firewall.domain.inbound | Domain inbound | High | allow\n- firewall.private.enabled | Private firewall | High | off\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\n- malware.realtime: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\n- firewall.domain.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-K5GG34O - Microsoft Windows 11 Pro 10.0.26100\", \"score\": {\"compliant\": 15, \"assessed\": 57, \"percent\": 26}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 42, \"NotApplicable\": 3, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"Microsoft Defender Antimalware\", \"risk\": \"High\", \"found\": \"service True, antivirus False\"}, {\"check_id\": \"malware.realtime\", \"name\": \"Real-time protection\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.enabled\", \"name\": \"Domain firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.private.enabled\", \"name\": \"Private firewall\", \"risk\": \"High\", \"found\": \"off\"}], \"summary\": \"LAPTOP-K5GG34O meets 15 of 57 assessed checks (26%). The most serious gaps are Microsoft Defender Antimalware, Real-time protection and Domain firewall. 4 items are pending.\", \"next_steps\": [\"Microsoft Defender Antimalware: Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\", \"Real-time protection: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\", \"Domain firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Answer the 4 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 3 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Tahoe 26.6.3-23J426, Command Line Tools for Xcode-16.4, Safari18.6Tahoe-18.6.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Found: 3 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 15.6 (macos)\nHOSTNAME: MKT-MacBook-57\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=32, Non-Compliant=15, NotApplicable=4, Pending=5, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 12\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5219\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5339, 17 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 60 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (asmith)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no recovery key\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 101 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 9 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 15.6\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Non-Compliant | Network Time: Off\n- gdpr.backup_configured | Backup configured | Compliant | bzserv\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | filesharing\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Hugging Face models, Perplexity\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Hugging Face models, Perplexity\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Hugging Face models, Perplexity\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.complexity: Medium\n- lockout.duration: Low\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.time_sync: Low\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.ai_literacy: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=32, assessed=47, percent=68\nTOP GAPS (computed, highest risk first):\n- user.account_type | Signed-in user account type | High | Administrator (asmith)\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 101 critical/high\n- gdpr.remote_access | Remote access services restricted | High | filesharing\n- org.ropa | Record of processing activities | High | no\n- password.min_length | Minimum password length | Medium | 12\nFIX GUIDANCE (highest-risk gaps):\n- user.account_type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.remote_access: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host)."}, {"role": "assistant", "content": "{\"device\": \"MKT-MacBook-57 - macOS 15.6\", \"score\": {\"compliant\": 32, \"assessed\": 47, \"percent\": 68}, \"counts\": {\"Compliant\": 32, \"Non-Compliant\": 15, \"NotApplicable\": 4, \"Pending\": 5, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (asmith)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"101 critical/high\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"filesharing\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"no\"}, {\"check_id\": \"password.min_length\", \"name\": \"Minimum password length\", \"risk\": \"Medium\", \"found\": \"12\"}], \"summary\": \"MKT-MacBook-57 meets 32 of 47 assessed checks (68%). The most serious gaps are Signed-in user account type, Known vulnerabilities in installed software and Remote access services restricted. 5 items are pending.\", \"next_steps\": [\"Signed-in user account type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Remote access services restricted: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.7.1 (ios)\nHOSTNAME: LEGAL-iPhone-06\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=8, Non-Compliant=11, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 0 detected by the browser, 21 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iOS 26.7.1 (up to date)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 14 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | off\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.7.1\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Perplexity\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Perplexity\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Perplexity\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | ChatGPT extension off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\nSCORE (computed): compliant=8, assessed=19, percent=42\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 14 critical/high\n- gdpr.auto_updates | Automatic iOS updates | High | off\n- gdpr.backup_configured | iCloud Backup | High | no backup\n- screen_lock.timeout | Auto-Lock | Medium | Auto-Lock Never\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.auto_updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files."}, {"role": "assistant", "content": "{\"device\": \"LEGAL-iPhone-06 - iOS 26.7.1\", \"score\": {\"compliant\": 8, \"assessed\": 19, \"percent\": 42}, \"counts\": {\"Compliant\": 8, \"Non-Compliant\": 11, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"14 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Auto-Lock\", \"risk\": \"Medium\", \"found\": \"Auto-Lock Never\"}], \"summary\": \"LEGAL-iPhone-06 meets 8 of 19 assessed checks (42%). The most serious gaps are Apps only from the App Store, Known vulnerabilities in iOS and Automatic iOS updates. 10 items are pending. 21 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Automatic iOS updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Found: no. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Real-time protection [malware.realtime] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Real-time protection scans files and processes as they are used.\nrisk: High - Without it, malware is only found by scheduled scans, after it has run.\nfix (windows):\n- Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\nverify: Re-run the audit; Real-time protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.realtime\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Real-time protection\", \"finding\": \"The device reports off, but the baseline requires on. Real-time protection scans files and processes as they are used.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without it, malware is only found by scheduled scans, after it has run.\"}, \"recommendation\": [\"Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Real-time protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.realtime", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Minimum password length is 6 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: screen lock set\nREQUIRED: PIN, password or pattern\nDETAIL: The browser reports that a device unlock method (passcode, PIN, Face ID or fingerprint) is set up.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock is compliant\", \"finding\": \"Found: screen lock set. This satisfies the baseline requirement of PIN, password or pattern.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Real-time protection [malware.realtime] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Real-time protection scans files and processes as they are used.\nrisk: High - Without it, malware is only found by scheduled scans, after it has run.\nfix (windows):\n- Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\nverify: Re-run the audit; Real-time protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.realtime\", \"status\": \"Non-Compliant\", \"title\": \"Real-time protection is non-compliant\", \"finding\": \"The device reports off, but the baseline requires on. Real-time protection scans files and processes as they are used.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without it, malware is only found by scheduled scans, after it has run.\"}, \"recommendation\": [\"Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Real-time protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.realtime", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength is non-compliant\", \"finding\": \"The user reports 4-digit code, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Found: USB drives writable without encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Diagnostic data minimised is not configured instead of required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted: does not meet the baseline\", \"finding\": \"Found: not restricted. Required: restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: off\nREQUIRED: off\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Compliant\", \"title\": \"USB debugging off: meets the baseline\", \"finding\": \"USB debugging off is set to off, in line with the required off. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 42 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.6 checked against NVD iOS records. iOS 26.6 (42 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-28935). 114 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS is non-compliant\", \"finding\": \"Known vulnerabilities in iOS is 42 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Found: partial. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 95 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 782 packages checked against rhel:9 advisories. flatpak-debugsource 0:1.12.9-3.el9_5 (2 CVEs, critical; fixed in 0:1.12.9-4.el9_8.1; e.g. CVE-2026-34078, CVE-2026-34079); java-17-openjdk-slowdebug-debuginfo 1:17.0.8.0.7-2.el9 (50 CVEs, high; fixed in 1:17.0.20.0.8-1.2.el9; e.g. CVE-2023-48161, CVE-2024-21208); postgresql-static 0:16.11-1.module+el9.7.0+23784+0c5a3b34 (30 CVEs, high; fixed in 0:16.14-1.module+el9.8.0+24360+ff98e86a; e.g. CVE-2026-14662, CVE-2026-14664); ghostscript-debugsource 0:9.54.0-16.el9_4 (6 CVEs, high; fixed in 0:9.54.0-18.el9_6; e.g. CVE-2023-46751, CVE-2024-46951); libreoffice-pyuno 1:7.1.8.1-14.el9_4 (3 CVEs, high; fixed in 1:7.1.8.1-15.el9_8.2; e.g. CVE-2025-1080, CVE-2026-4430); and 3 more. 21 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"Found: 95 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iPadOS 18.7.8 (ios)\nAI TOOLS FOUND: Gemini\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Gemini.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 10\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Chrome 138). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"The device reports Android 10, but the baseline requires >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Private outbound\", \"finding\": \"Found: block. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training matches the required setting\", \"finding\": \"AI literacy training is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Usage & diagnostics sharing minimised\", \"finding\": \"Usage & diagnostics sharing minimised is usage & diagnostics on instead of usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pattern\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: Pattern locks are easy to guess from screen smudges or by watching.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength is non-compliant\", \"finding\": \"The user reports pattern, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Hugging Face models\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Hugging Face models. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Removable media encryption is external drives writable without enforced encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"Found: not set. Required: >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: passcode set\nREQUIRED: passcode set\nDETAIL: The browser reports that a device unlock method (passcode, PIN, Face ID or fingerprint) is set up.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Passcode: meets the baseline\", \"finding\": \"Found: passcode set. This satisfies the baseline requirement of passcode set.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: DEV-MacBook-39\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=43, Non-Compliant=5, NotApplicable=4, Pending=3, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5343\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5280, 13 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Error | not read\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 65 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 0 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (mkhan)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 11 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | ArqAgent\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | remotemanagement\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Ollama, Perplexity, Jan\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Ollama, Perplexity, Jan\n- aiact.risk_classification | AI system risk classification | Pending | Ollama, Perplexity, Jan\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Intel Mac\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- usb.write: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- org.processor_agreements: High\nSCORE (computed): compliant=43, assessed=48, percent=90\nTOP GAPS (computed, highest risk first):\n- gdpr.remote_access | Remote access services restricted | High | remotemanagement\n- org.processor_agreements | Data processing agreements | High | partial\n- usb.write | USB storage write | Medium | allow\n- gdpr.removable_encryption | Removable media encryption | Medium | external drives writable without enforced encryption\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | Ollama, Perplexity, Jan\nFIX GUIDANCE (highest-risk gaps):\n- gdpr.remote_access: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\n- org.processor_agreements: List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\n- usb.write: Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product."}, {"role": "assistant", "content": "{\"device\": \"DEV-MacBook-39 - macOS 26.0.1\", \"score\": {\"compliant\": 43, \"assessed\": 48, \"percent\": 90}, \"counts\": {\"Compliant\": 43, \"Non-Compliant\": 5, \"NotApplicable\": 4, \"Pending\": 3, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"remotemanagement\"}, {\"check_id\": \"org.processor_agreements\", \"name\": \"Data processing agreements\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"usb.write\", \"name\": \"USB storage write\", \"risk\": \"Medium\", \"found\": \"allow\"}, {\"check_id\": \"gdpr.removable_encryption\", \"name\": \"Removable media encryption\", \"risk\": \"Medium\", \"found\": \"external drives writable without enforced encryption\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"Ollama, Perplexity, Jan\"}], \"summary\": \"DEV-MacBook-39 meets 43 of 48 assessed checks (90%). The most serious gaps are Remote access services restricted and Data processing agreements. 1 checks could not be read. 3 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Remote access services restricted: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\", \"Data processing agreements: List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\", \"USB storage write: Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"Found: no. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 203 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 18 applications inventoried, 10 recognised (23 common applications covered). Firefox 152.3 (201 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100768, CVE-2026-100779); VLC 3.0.19 (2 CVEs, critical; fixed in 3.0.20; e.g. CVE-2023-47359, CVE-2023-47360). 38 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 203 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 61 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 23 applications inventoried, 12 recognised (31 common applications covered). Adobe Acrobat Reader (64-bit) 24.005.20320 (55 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-27158, CVE-2025-27159); Microsoft Teams 1.6.00.26474 (6 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330). 34 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"The device reports 61 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"Password history is 5 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with pinHistory = 24.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"Data protection impact assessment is no instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption protection is non-compliant\", \"finding\": \"Found: not encrypted. Required: active, strong cipher. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: usb-storage module disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Non-Compliant\", \"title\": \"USB storage device access is non-compliant\", \"finding\": \"The device reports deny, but the baseline requires allow. Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The device access setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage device access should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Jan, GPT4All\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Jan, GPT4All.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Android 13\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Firefox 143).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version: meets the baseline\", \"finding\": \"Found: Android 13. This satisfies the baseline requirement of >= Android 13.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"ClamAV malware protection service: does not meet the baseline\", \"finding\": \"ClamAV malware protection service is not running instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"Found: no. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Passcode strength\", \"finding\": \"Passcode strength is 4-digit code instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.4 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates is non-compliant\", \"finding\": \"iOS security updates is iOS 26.4 (update to 26.7.1 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nHOSTNAME: DESKTOP-4215OE9\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=37, Non-Compliant=19, NotApplicable=1, Pending=6, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 10\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Compliant | on\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Non-Compliant | off\n- malware.signatures | Security intelligence | Non-Compliant | 19 days old (version 1.440.646.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Non-Compliant | block\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Compliant | 300 seconds (machine inactivity limit)\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | DESKTOP-4215OE9\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-4215OE9\\asmith\n- user.account_type | Signed-in user account type | Compliant | Standard (DESKTOP-4215OE9\\tnguyen)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 1 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 61 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | AUOptions=4\n- gdpr.updates_recent | Security updates installed recently | Compliant | 19 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 10 Pro build 19045\n- gdpr.security_logging | Security event logging | Error | not read\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude, Microsoft Copilot\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude, Microsoft Copilot\n- aiact.risk_classification | AI system risk classification | Pending | Claude, Microsoft Copilot\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Compliant | Recall disabled by policy\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.min_age: Low\n- password.complexity: Medium\n- malware.tamper: Medium\n- malware.signatures: High\n- firewall.public.outbound: Low\n- usb.write: Medium\n- admin.unauthorized: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=37, assessed=56, percent=66\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Security intelligence | High | 19 days old (version 1.440.646.0)\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | DESKTOP-4215OE9\\asmith\n- vuln.pending_updates | Pending security updates | High | 2 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 61 critical/high\n- gdpr.os_supported | Supported operating system version | High | Microsoft Windows 10 Pro build 19045\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: Run Update-MpSignature and check that the device can reach Windows Update or your update server.\n- admin.unauthorized: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\n- vuln.pending_updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-4215OE9 - Microsoft Windows 10 Pro 10.0.19045\", \"score\": {\"compliant\": 37, \"assessed\": 56, \"percent\": 66}, \"counts\": {\"Compliant\": 37, \"Non-Compliant\": 19, \"NotApplicable\": 1, \"Pending\": 6, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Security intelligence\", \"risk\": \"High\", \"found\": \"19 days old (version 1.440.646.0)\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"DESKTOP-4215OE9\\\\asmith\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"2 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"61 critical/high\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Microsoft Windows 10 Pro build 19045\"}], \"summary\": \"DESKTOP-4215OE9 meets 37 of 56 assessed checks (66%). The most serious gaps are Security intelligence, Unauthorized employee local administrator accounts and Pending security updates. 1 checks could not be read. 6 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Security intelligence: Run Update-MpSignature and check that the device can reach Windows Update or your update server.\", \"Unauthorized employee local administrator accounts: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\", \"Pending security updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\", \"Answer the 4 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Found: Claude Code. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 14.7.8 (macos)\nHOSTNAME: DEV-MacBook-19\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=40, Non-Compliant=10, NotApplicable=5, Pending=1, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5237; third-party: Bitdefender Endpoint Security\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5393, 20 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | Compliant | third-party scanner: Bitdefender Endpoint Security\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall on, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 1200 idle, password delay 60\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 0 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (sali)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 94 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 4 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 14.7.8\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | restricted by profile\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.complexity: Medium\n- lockout.duration: Low\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- org.breach_procedure: High\nSCORE (computed): compliant=40, assessed=50, percent=80\nTOP GAPS (computed, highest risk first):\n- firewall.inbound | Firewall inbound policy | High | firewall on, stealth mode off\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 94 critical/high\n- gdpr.backup_configured | Backup configured | High | none\n- org.breach_procedure | Breach response procedure | High | partial\n- password.complexity | Password complexity | Medium | not enforced\nFIX GUIDANCE (highest-risk gaps):\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.backup_configured: Configure Time Machine (tmutil setdestination) or the organisation's backup agent."}, {"role": "assistant", "content": "{\"device\": \"DEV-MacBook-19 - macOS 14.7.8\", \"score\": {\"compliant\": 40, \"assessed\": 50, \"percent\": 80}, \"counts\": {\"Compliant\": 40, \"Non-Compliant\": 10, \"NotApplicable\": 5, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall on, stealth mode off\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"94 critical/high\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"org.breach_procedure\", \"name\": \"Breach response procedure\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"password.complexity\", \"name\": \"Password complexity\", \"risk\": \"Medium\", \"found\": \"not enforced\"}], \"summary\": \"DEV-MacBook-19 meets 40 of 50 assessed checks (80%). The most serious gaps are Firewall inbound policy, Known vulnerabilities in installed software and Backup configured. 1 items are pending.\", \"next_steps\": [\"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Backup configured: Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: popularity-contest on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"The device reports popularity-contest on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Compliant\", \"title\": \"Record of processing activities is compliant\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nHOSTNAME: legal-lt-30\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=38, Non-Compliant=13, NotApplicable=4, Pending=1, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Non-Compliant | PASS_MIN_DAYS 0\n- password.complexity | Password complexity | Compliant | minclass=0, required classes=4\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 1 days old\n- malware.on_access | On-access malware protection | Compliant | clamonacc running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: akaya (1800s)\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (akaya)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 3 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 4 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 26.04.1 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.min_age: Low\n- lockout.observation: Low\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- screen_lock.timeout: Medium\n- admin.approved: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.telemetry: Low\n- org.dpia: Medium\n- org.ai_register: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=38, assessed=51, percent=75\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\n- gdpr.backup_configured | Backup configured | High | none\n- screen_lock.timeout | Screen lock timeout | Medium | too long for: akaya (1800s)\n- admin.approved | Approved IT administrator | Medium | none of the approved accounts is an administrator\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\n- firewall.inbound: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\n- gdpr.backup_configured: Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup)."}, {"role": "assistant", "content": "{\"device\": \"legal-lt-30 - Ubuntu 26.04.1 LTS 26.04\", \"score\": {\"compliant\": 38, \"assessed\": 51, \"percent\": 75}, \"counts\": {\"Compliant\": 38, \"Non-Compliant\": 13, \"NotApplicable\": 4, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Screen lock timeout\", \"risk\": \"Medium\", \"found\": \"too long for: akaya (1800s)\"}, {\"check_id\": \"admin.approved\", \"name\": \"Approved IT administrator\", \"risk\": \"Medium\", \"found\": \"none of the approved accounts is an administrator\"}], \"summary\": \"legal-lt-30 meets 38 of 51 assessed checks (75%). The most serious gaps are Firewall, Firewall inbound policy and Backup configured. 1 items are pending.\", \"next_steps\": [\"Firewall: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\", \"Firewall inbound policy: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\", \"Backup configured: Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: FileVault off\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Encryption\", \"finding\": \"Encryption is FileVault off instead of FileVault on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"The device reports share Mac analytics=1, share with app developers=0, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"Human oversight for high-risk AI is no instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"Password history is 0 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates: does not meet the baseline\", \"finding\": \"The user reports automatic app updates off, but the baseline requires automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (macos):\n- Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"Approved IT administrator is none of the approved accounts is an administrator instead of itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Breach response procedure is no instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"The device reports none found, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"Diagnostic data minimised is share Mac analytics=1, share with app developers=0 instead of analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an enabled administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"The device reports none of the approved accounts is an enabled administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery: does not meet the baseline\", \"finding\": \"Found: key protectors: Tpm. Required: recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Microsoft Visual Studio Code (User) Microsoft.VisualStudioCode 100.3.739 100.3.7 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"Found: 1 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no recovery key\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"The device reports no recovery key, but the baseline requires recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-10 Cumulative Update for Windows 11 Version 24H2 for x64-based Systems (KB5049136), 2026-10 Cumulative Update for .NET Framework 3.5 and 4.8.1 for Windows 11 Version 24H2 for x64 (KB5046962).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Pending security updates is 2 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Sensitive notifications hidden when locked is non-compliant\", \"finding\": \"Found: all content shown. Required: sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (CONTOSO\\akaya)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Found: Administrator (CONTOSO\\\\akaya). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"Password complexity is disabled instead of enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Date & time set automatically on iPhone: Date & time set automatically on iPhone. Why it matters: Accurate clocks are needed to put events in the right order during an investigation. Risk if missing: Low - Wrong timestamps make incident timelines unreliable. Requirement: The clock is synchronised from a time server. How to fix on iPhone: (1) Settings › General › Date & Time: turn on Set Automatically. How to verify: Re-run the audit; Time synchronisation should be Compliant. Related controls: ISO 27001:2022 A.8.17 (Clock synchronization); GDPR Art. 33 (Notification of a personal data breach to the supervisory authority).\n[2] Where to find Date & time set automatically on Android: Date & time set automatically on Android: Is 'Set time automatically' on? Where to look: Settings › System › Date & time\n[3] Date & time set automatically on Android: Date & time set automatically on Android. Why it matters: Accurate clocks are needed to put events in the right order during an investigation. Risk if missing: Low - Wrong timestamps make incident timelines unreliable. Requirement: The clock is synchronised from a time server. How to fix on Android: (1) Settings › System › Date & time: turn on Set time automatically. How to verify: Re-run the audit; Time synchronisation should be Compliant. Related controls: ISO 27001:2022 A.8.17 (Clock synchronization); GDPR Art. 33 (Notification of a personal data breach to the supervisory authority).\n[4] Where to find Date & time set automatically on iPhone: Date & time set automatically on iPhone: Is 'Set Automatically' on? Where to look: Settings › General › Date & Time\n\nMY AUDIT: none\n\nMY DEVICE: iOS 17.6 (ios) on an iPhone, Safari 18.6; detected by the browser, security settings not checked.\n\nQUESTION: Why does date & time set automatically matter?"}, {"role": "assistant", "content": "Accurate clocks are needed to put events in the right order during an investigation. If it is missing, the risk is low: Wrong timestamps make incident timelines unreliable. [1]"}], "meta": {"task": "chat", "os": "ios", "status": "why", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (sali)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Signed-in user account type is Administrator (sali) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: check, download and install automatic\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates matches the required setting\", \"finding\": \"Automatic security updates is set to check, download and install automatic, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Block or allow external storage with your MDM (Restrictions / device control) or an endpoint DLP product.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Network Time: Off\nREQUIRED: network time on\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"Time synchronisation is Network Time: Off instead of network time on. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"sudo systemsetup -setusingnetworktime on.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 13 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-07 Cumulative Update for Windows 11 (2026-09-23)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Compliant\", \"title\": \"Security updates installed recently matches the required setting\", \"finding\": \"Security updates installed recently is set to 13 days ago, in line with the required <= 30 days ago.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.7 (ios)\nHOSTNAME: SALES-iPhone-16\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=6, Non-Compliant=13, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 3 detected by the browser, 18 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Non-Compliant | no passcode\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Non-Compliant | off (no passcode)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.7 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 15 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | ChatGPT\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- encryption.enabled: High\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=6, assessed=19, percent=32\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- encryption.enabled | Data protection (encryption) | High | off (no passcode)\n- vuln.pending_updates | iOS security updates | High | iOS 26.7 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 15 critical/high\n- gdpr.backup_configured | iCloud Backup | High | no backup\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- encryption.enabled: Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered."}, {"role": "assistant", "content": "{\"device\": \"SALES-iPhone-16 - iOS 26.7\", \"score\": {\"compliant\": 6, \"assessed\": 19, \"percent\": 32}, \"counts\": {\"Compliant\": 6, \"Non-Compliant\": 13, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Data protection (encryption)\", \"risk\": \"High\", \"found\": \"off (no passcode)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.7 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"15 critical/high\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}], \"summary\": \"SALES-iPhone-16 meets 6 of 19 assessed checks (32%). The most serious gaps are Apps only from the App Store, Data protection (encryption) and iOS security updates. 10 items are pending. 18 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Data protection (encryption): Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Compliant\", \"title\": \"AI register: meets the baseline\", \"finding\": \"AI register is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: macOS 15.7.2\nREQUIRED: macOS 14 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (macos):\n- Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version is compliant\", \"finding\": \"Found: macOS 15.7.2. This satisfies the baseline requirement of macOS 14 or later.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"Found: share Mac analytics=1, share with app developers=0. Required: analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 18 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 20 applications inventoried, 10 recognised (23 common applications covered). VLC 3.0.12 (4 CVEs, critical; fixed in 3.0.20; e.g. CVE-2022-41325, CVE-2023-46814); Safari 26.4 (14 CVEs, high; fixed in 26.6.1; e.g. CVE-2026-43670, CVE-2026-43701). 50 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 18 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: usage reporting off\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Compliant\", \"title\": \"Diagnostic data minimised is compliant\", \"finding\": \"Found: usage reporting off. This satisfies the baseline requirement of usage reporting off.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: backed up to a computer\nREQUIRED: backup on\nDETAIL: Manual computer backups depend on the user remembering them.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Compliant\", \"title\": \"iCloud Backup: meets the baseline\", \"finding\": \"The user reports backed up to a computer, which meets the requirement (backup on). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: none\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Error\", \"title\": \"Unauthorized employee administrator accounts: evidence missing\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Domain inbound\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAI TOOLS FOUND: LM Studio, Claude Code, Hugging Face models\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /home/jdoe/.cache/huggingface/hub (14G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- LM Studio: Local model runner; provider Element Labs (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences and whether its local server is exposed.\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"LM Studio\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences and whether its local server is exposed.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: LM Studio, Claude Code, Hugging Face models.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history: does not meet the baseline\", \"finding\": \"The device reports 0, but the baseline requires >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Grok\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: Grok. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Gemini, Grammarly\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: ChatGPT, Gemini, Grammarly. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Found: no machine inactivity limit; screen saver off. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 pending\nREQUIRED: <= 0 pending\nDETAIL: Uses the cached repository data.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Compliant\", \"title\": \"Pending security updates matches the required setting\", \"finding\": \"Pending security updates is set to 0 pending, in line with the required <= 0 pending.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public inbound\", \"finding\": \"Public inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 15.7.2 (macos)\nHOSTNAME: OPS-MacBook-55\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=19, Non-Compliant=28, NotApplicable=3, Pending=5, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Non-Compliant | not set\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5392\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5309, 10 days old\n- malware.signature_updates | Malware security updates | Non-Compliant | ConfigDataInstall=0, CriticalUpdateInstall=1\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 600 idle, password delay 60\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | tnguyen, ewagner\n- user.account_type | Signed-in user account type | Compliant | Standard (asmith)\n- encryption.enabled | Encryption | Non-Compliant | FileVault off\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no FileVault\n- vuln.pending_updates | Pending security updates | Non-Compliant | 3 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 61 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 38 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 15.7.2\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Non-Compliant | Network Time: Off\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | filesharing, anydesk\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | LM Studio, Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Error | not read\n- aiact.risk_classification | AI system risk classification | Pending | LM Studio, Claude Code\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.observation: Low\n- malware.signature_updates: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=19, assessed=47, percent=40\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | tnguyen, ewagner\n- encryption.enabled | Encryption | High | FileVault off\n- encryption.protection | Disk encryption protection | High | not encrypted\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin."}, {"role": "assistant", "content": "{\"device\": \"OPS-MacBook-55 - macOS 15.7.2\", \"score\": {\"compliant\": 19, \"assessed\": 47, \"percent\": 40}, \"counts\": {\"Compliant\": 19, \"Non-Compliant\": 28, \"NotApplicable\": 3, \"Pending\": 5, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"tnguyen, ewagner\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"FileVault off\"}, {\"check_id\": \"encryption.protection\", \"name\": \"Disk encryption protection\", \"risk\": \"High\", \"found\": \"not encrypted\"}], \"summary\": \"OPS-MacBook-55 meets 19 of 47 assessed checks (40%). The most serious gaps are Firewall, Firewall inbound policy and Unauthorized employee administrator accounts. 1 checks could not be read. 5 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Answer the 3 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 570 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 20 applications inventoried, 12 recognised (31 common applications covered). Google Chrome 151.0.7922.173 (402 CVEs, critical, actively exploited; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301); Adobe Acrobat Reader (64-bit) 25.001.20531 (43 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-54257, CVE-2025-64785); Mozilla Firefox (x64 en-US) 153.1.0 (125 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100756, CVE-2026-100767). 578 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 570 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: enabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Guest account disabled\", \"finding\": \"The device reports enabled, but the baseline requires disabled. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted is non-compliant\", \"finding\": \"The device reports not restricted, but the baseline requires restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates: does not meet the baseline\", \"finding\": \"Automatic system updates is manual instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold is non-compliant\", \"finding\": \"The device reports not set, but the baseline requires 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxFailedAttempts = 5.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pmartin\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: pmartin. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rooted or bootloader unlocked\nREQUIRED: not rooted\nDETAIL: Android security protections are weakened on a rooted phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not rooted is non-compliant\", \"finding\": \"Not rooted is rooted or bootloader unlocked instead of not rooted. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Hugging Face models\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Hugging Face models.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answer ORG_DPIA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: download and install on\nDETAIL: Security updates are not installed automatically.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic iOS updates\", \"finding\": \"Automatic iOS updates is off instead of download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 14.7.8 (macos)\nHOSTNAME: MKT-MacBook-37\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=31, Non-Compliant=17, NotApplicable=3, Pending=5, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Non-Compliant | not set\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5340\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5391, 14 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall on, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 180 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after 300 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | rschmit\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (sali)\n- encryption.enabled | Encryption | Non-Compliant | FileVault off\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no FileVault\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 6 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 15 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 14.7.8\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | LM Studio\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | LM Studio\n- aiact.risk_classification | AI system risk classification | Pending | LM Studio\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.history: Low\n- lockout.observation: Low\n- firewall.inbound: High\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\nSCORE (computed): compliant=31, assessed=48, percent=65\nTOP GAPS (computed, highest risk first):\n- firewall.inbound | Firewall inbound policy | High | firewall on, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | rschmit\n- user.account_type | Signed-in user account type | High | Administrator (sali)\n- encryption.enabled | Encryption | High | FileVault off\n- encryption.protection | Disk encryption protection | High | not encrypted\nFIX GUIDANCE (highest-risk gaps):\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\n- user.account_type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account."}, {"role": "assistant", "content": "{\"device\": \"MKT-MacBook-37 - macOS 14.7.8\", \"score\": {\"compliant\": 31, \"assessed\": 48, \"percent\": 65}, \"counts\": {\"Compliant\": 31, \"Non-Compliant\": 17, \"NotApplicable\": 3, \"Pending\": 5, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall on, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"rschmit\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (sali)\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"FileVault off\"}, {\"check_id\": \"encryption.protection\", \"name\": \"Disk encryption protection\", \"risk\": \"High\", \"found\": \"not encrypted\"}], \"summary\": \"MKT-MacBook-37 meets 31 of 48 assessed checks (65%). The most serious gaps are Firewall inbound policy, Unauthorized employee administrator accounts and Signed-in user account type. 5 items are pending.\", \"next_steps\": [\"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Signed-in user account type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\", \"Answer the 3 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 12\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Chrome 141). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"Supported operating system version is Android 12 instead of >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates is non-compliant\", \"finding\": \"App updates is automatic app updates off instead of automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iOS 26.4\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Safari 26.4).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version matches the required setting\", \"finding\": \"The device reports iOS 26.4, which meets the requirement (>= iOS 18).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 46 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 45; flatpaks: us.zoom.Zoom\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Outdated third-party applications is 46 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"The device reports 5 minutes, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutwindow:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: chromeremotedesktop, filesharing\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"The device reports chromeremotedesktop, filesharing, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password age\", \"finding\": \"Found: 0 days. Required: >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery: does not meet the baseline\", \"finding\": \"Found: key protectors: Tpm. Required: recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"Approved IT administrator is none of the approved accounts is an administrator instead of itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answer ORG_AI_TRANSPARENCY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 3 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Git Git.Git 89.5.630 90.9.896 winget; 7-Zip 24.08 (x64) 7zip.7zip 1.2.772 1.3.901 winget; Mozilla Firefox (x64 en-US) Mozilla.Firefox 6.1.591 6.6.732 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Outdated third-party applications is 3 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private firewall [firewall.private.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Private firewall\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 31 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: libxml2, kernel, glibc-common. Uses the cached repository data.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Found: 31 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: lmuller\nREQUIRED: password required\nDETAIL: Users checked: lmuller\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"Found: not required for: lmuller. Required: password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 600 idle, password delay 0\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"Found: 600 idle, password delay 0. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall\", \"finding\": \"Firewall is no active firewall instead of active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption recovery\", \"finding\": \"The device reports not encrypted, but the baseline requires recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"AI literacy training is partial instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: popularity-contest on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"The device reports popularity-contest on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: all removable storage denied by policy\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (windows):\n- Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Non-Compliant\", \"title\": \"USB storage read: does not meet the baseline\", \"finding\": \"Found: deny. Required: allow. The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage read should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 222 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 23 applications inventoried, 14 recognised (31 common applications covered). Microsoft Teams 1.6.00.18681 (7 CVEs, high, actively exploited; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330); TeamViewer 13.2.36216 (4 CVEs, high, actively exploited; fixed in 15.21.2; e.g. CVE-2019-18988, CVE-2020-13699); Google Chrome 153.0.8010.52 (137 CVEs, critical; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301); Microsoft Visual Studio Code (User) 1.123.2 (19 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-50520, CVE-2026-57102); Zoom Workplace 6.4.13 (8 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-49460, CVE-2025-49461); and 4 more. 186 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"Known vulnerabilities in installed software is 222 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 10\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Samsung Internet 28.0). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"Supported operating system version is Android 10 instead of >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nDETAIL: Users checked: ewagner\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock: meets the baseline\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Windows Defender (off, out of date)\nREQUIRED: enabled and up to date\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (windows):\n- Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"The device reports Windows Defender (off, out of date), but the baseline requires enabled and up to date. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"Found: none found. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 10\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Firefox 143). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"Supported operating system version is Android 10 instead of >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Red Hat Enterprise Linux 8.10 (Ootpa)\nREQUIRED: RHEL 9 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"The device reports Red Hat Enterprise Linux 8.10 (Ootpa), but the baseline requires RHEL 9 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAI TOOLS FOUND: Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period matches the required setting\", \"finding\": \"Lockout observation period is set to 30 minutes, in line with the required >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Ollama\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: Ollama. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: fully encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"Disk encryption protection is compliant\", \"finding\": \"The device reports fully encrypted, which meets the requirement (fully encrypted).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period is compliant\", \"finding\": \"The device reports 30 minutes, which meets the requirement (>= 15 minutes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"The user reports assistant screen access on, but the baseline requires screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 35 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-08-30 01:19:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Found: 35 days ago. Required: <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MIN_DAYS 0\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (linux):\n- Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age is non-compliant\", \"finding\": \"The device reports PASS_MIN_DAYS 0, but the baseline requires >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"The device reports 10, but the baseline requires >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\", \"For a stand-alone device: net accounts /uniquepw:24\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: unknown apps blocked for all apps\nREQUIRED: all apps: Not allowed\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Compliant\", \"title\": \"Install unknown apps blocked: meets the baseline\", \"finding\": \"Install unknown apps blocked is set to unknown apps blocked for all apps, in line with the required all apps: Not allowed. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Human oversight for high-risk AI is partial instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect 5213, 22 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Compliant\", \"title\": \"Malware protection signatures is compliant\", \"finding\": \"Found: XProtect 5213, 22 days old. This satisfies the baseline requirement of <= 30 days old.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: screensharing\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Found: screensharing. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: writing to USB storage blocked\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Compliant\", \"title\": \"Removable media encryption is compliant\", \"finding\": \"The device reports writing to USB storage blocked, which meets the requirement (no unencrypted writes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (CONTOSO\\jdoe)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"Signed-in user account type is Administrator (CONTOSO\\\\jdoe) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode on\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"The device reports firewall off, stealth mode on, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: OPS-MacBook-58\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=15, Non-Compliant=31, NotApplicable=6, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Non-Compliant | not set\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout\n- malware.antivirus | Antivirus | Compliant | XProtect 5360\n- malware.service | Apple malware protection service | Non-Compliant | XProtect scanning on; Gatekeeper assessments disabled\n- malware.signatures | Malware protection signatures | Non-Compliant | XProtect 5389, 104 days old\n- malware.signature_updates | Malware security updates | Non-Compliant | ConfigDataInstall=0, CriticalUpdateInstall=1\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 60 idle, password delay 300\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | rschmit, tnguyen\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (ewagner)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no recovery key\n- vuln.pending_updates | Pending security updates | Non-Compliant | 3 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 37 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 314 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 75 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Non-Compliant | Network Time: Off\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Non-Compliant | guest login 1, guest file sharing 0\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | teamviewer\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Intel Mac\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.expiration: Low\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.service: High\n- malware.signatures: High\n- malware.signature_updates: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=15, assessed=46, percent=33\nTOP GAPS (computed, highest risk first):\n- malware.service | Apple malware protection service | High | XProtect scanning on; Gatekeeper assessments disabled\n- malware.signatures | Malware protection signatures | High | XProtect 5389, 104 days old\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | rschmit, tnguyen\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\n- malware.signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile."}, {"role": "assistant", "content": "{\"device\": \"OPS-MacBook-58 - macOS 26.0.1\", \"score\": {\"compliant\": 15, \"assessed\": 46, \"percent\": 33}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 31, \"NotApplicable\": 6, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"Apple malware protection service\", \"risk\": \"High\", \"found\": \"XProtect scanning on; Gatekeeper assessments disabled\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware protection signatures\", \"risk\": \"High\", \"found\": \"XProtect 5389, 104 days old\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"rschmit, tnguyen\"}], \"summary\": \"OPS-MacBook-58 meets 15 of 46 assessed checks (33%). The most serious gaps are Apple malware protection service, Malware protection signatures and Firewall. 4 items are pending.\", \"next_steps\": [\"Apple malware protection service: Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\", \"Malware protection signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\", \"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Answer the 4 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Password complexity is disabled instead of enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: guest login off\nREQUIRED: guest login off\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (macos):\n- sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled is compliant\", \"finding\": \"Found: guest login off. This satisfies the baseline requirement of guest login off.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history: does not meet the baseline\", \"finding\": \"Password history is 0 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: XProtect 5272, 61 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware protection signatures: does not meet the baseline\", \"finding\": \"Found: XProtect 5272, 61 days old. Required: <= 30 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; no signed-in user to check\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock is non-compliant\", \"finding\": \"Screen lock is no machine inactivity limit; no signed-in user to check instead of on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rschmit\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"Found: rschmit. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"The device reports none, which meets the requirement (only: none).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 26.0.1 (macos)\nAI TOOLS FOUND: Jan\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Jan: Local model runner; provider Open source (runs locally); data: On the device unless a cloud model is connected; default risk minimal; GDPR: Check whether remote API providers are configured; if so, the cloud provider rules apply.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Jan\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Check whether remote API providers are configured; if so, the cloud provider rules apply.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Jan.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: <= 0 pending\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Error\", \"title\": \"Pending security updates: could not be checked\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MIN_DAYS 0\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (linux):\n- Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"The device reports PASS_MIN_DAYS 0, but the baseline requires >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 10 Pro build 19045\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"Supported operating system version is Microsoft Windows 10 Pro build 19045 instead of build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"The device reports 10, but the baseline requires 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxFailedAttempts = 5.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"Found: partial. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Password complexity is disabled instead of enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: restic-backup.timer.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Confirm the last backup date in: restic-backup.timer. Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Hugging Face models\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Unapproved AI tools (shadow AI) is Hugging Face models instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"Found: partial. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no screen lock (Swipe/None)\nREQUIRED: PIN, password or pattern\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Found: no screen lock (Swipe/None). Required: PIN, password or pattern. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: no such setting\nREQUIRED: >= 1 days\nDETAIL: macOS account policies have no minimum password age.\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (macos):\n- macOS has no minimum password age setting; rely on password history instead.\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"NotApplicable\", \"title\": \"Minimum password age: not assessed on this device\", \"finding\": \"Not applicable on this device: macOS account policies have no minimum password age.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (asmith)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"The device reports Administrator (asmith), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 2 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Compliant\", \"title\": \"Recent backup: meets the baseline\", \"finding\": \"The device reports 2 days ago, which meets the requirement (<= 7 days ago).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: tnguyen\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Unauthorized employee administrator accounts is tnguyen instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: ChatGPT, Perplexity\nREQUIRED: inventory taken\nDETAIL: AI apps listed by the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (ios):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory is compliant\", \"finding\": \"The user reports ChatGPT, Perplexity, which meets the requirement (inventory taken). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: success; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security event logging\", \"finding\": \"Security event logging is logon: success; account management: none instead of logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: recovery key or second unlock method\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Error\", \"title\": \"Disk encryption recovery: evidence missing\", \"finding\": \"This check returned an error: Requires root: run the collector with sudo. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.7 (ios)\nHOSTNAME: SALES-iPhone-37\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=11, Non-Compliant=7, NotApplicable=2, Pending=9, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store only\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.7 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 15 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Pending | not answered\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.lock_screen_previews: Low\nSCORE (computed): compliant=11, assessed=18, percent=61\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | iOS security updates | High | iOS 26.7 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 15 critical/high\n- gdpr.backup_configured | iCloud Backup | High | no backup\n- screen_lock.timeout | Auto-Lock | Medium | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Medium | 4-digit code\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.backup_configured: Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps."}, {"role": "assistant", "content": "{\"device\": \"SALES-iPhone-37 - iOS 26.7\", \"score\": {\"compliant\": 11, \"assessed\": 18, \"percent\": 61}, \"counts\": {\"Compliant\": 11, \"Non-Compliant\": 7, \"NotApplicable\": 2, \"Pending\": 9, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.7 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"15 critical/high\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Auto-Lock\", \"risk\": \"Medium\", \"found\": \"Auto-Lock Never\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Passcode strength\", \"risk\": \"Medium\", \"found\": \"4-digit code\"}], \"summary\": \"SALES-iPhone-37 meets 11 of 18 assessed checks (61%). The most serious gaps are iOS security updates, Known vulnerabilities in iOS and iCloud Backup. 9 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"iCloud Backup: Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall inbound policy\", \"finding\": \"Firewall inbound policy is allow (no firewall) instead of deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"Lockout observation period is not set instead of >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 90 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration: meets the baseline\", \"finding\": \"Found: 90 days. This satisfies the baseline requirement of 1-365 days.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain inbound: does not meet the baseline\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nHOSTNAME: LAPTOP-XW6H966\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=46, Non-Compliant=5, NotApplicable=9, Pending=2, Error=2\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | Sophos Endpoint (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Error | not read\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Non-Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Error | not read\n- screen_lock.password | Screen lock password requirement | Compliant | machine inactivity limit locks the session\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | DEVSEIS\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (DEVSEIS\\pmartin)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | AUOptions=4\n- gdpr.updates_recent | Security updates installed recently | Compliant | 4 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 11 Pro build 22621\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | Veeam Agent for Microsoft Windows\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=0 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- firewall.domain.inbound: High\n- firewall.domain.outbound: Low\n- vuln.pending_updates: High\n- gdpr.os_supported: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=46, assessed=51, percent=90\nTOP GAPS (computed, highest risk first):\n- firewall.domain.inbound | Domain inbound | High | allow\n- vuln.pending_updates | Pending security updates | High | 1 pending\n- gdpr.os_supported | Supported operating system version | High | Microsoft Windows 11 Pro build 22621\n- org.ai_literacy | AI literacy training | Medium | partial\n- firewall.domain.outbound | Domain outbound | Low | allow\nFIX GUIDANCE (highest-risk gaps):\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- vuln.pending_updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\n- gdpr.os_supported: Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-XW6H966 - Microsoft Windows 11 Pro 10.0.22621\", \"score\": {\"compliant\": 46, \"assessed\": 51, \"percent\": 90}, \"counts\": {\"Compliant\": 46, \"Non-Compliant\": 5, \"NotApplicable\": 9, \"Pending\": 2, \"Error\": 2}, \"top_gaps\": [{\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"1 pending\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Microsoft Windows 11 Pro build 22621\"}, {\"check_id\": \"org.ai_literacy\", \"name\": \"AI literacy training\", \"risk\": \"Medium\", \"found\": \"partial\"}, {\"check_id\": \"firewall.domain.outbound\", \"name\": \"Domain outbound\", \"risk\": \"Low\", \"found\": \"allow\"}], \"summary\": \"LAPTOP-XW6H966 meets 46 of 51 assessed checks (90%). The most serious gaps are Domain inbound, Pending security updates and Supported operating system version. 2 checks could not be read. 2 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Pending security updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\", \"Supported operating system version: Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Network Time: Off\nREQUIRED: network time on\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation is non-compliant\", \"finding\": \"The device reports Network Time: Off, but the baseline requires network time on. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"sudo systemsetup -setusingnetworktime on.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history: does not meet the baseline\", \"finding\": \"The device reports not set, but the baseline requires >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with pinHistory = 24.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (macos):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery does not apply here\", \"finding\": \"This check does not apply: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Compliant\", \"title\": \"Public outbound is compliant\", \"finding\": \"Public outbound is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: enabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Compliant\", \"title\": \"Password complexity is compliant\", \"finding\": \"Found: enabled. This satisfies the baseline requirement of enabled.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ConfigDataInstall=0, CriticalUpdateInstall=1\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware security updates is non-compliant\", \"finding\": \"Found: ConfigDataInstall=0, CriticalUpdateInstall=1. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Domain firewall is non-compliant\", \"finding\": \"Domain firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Found: disabled. Required: enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: sali (0s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: sali, asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"The device reports too long for: sali (0s), but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (macos):\n- Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator is non-compliant\", \"finding\": \"Approved IT administrator is none of the approved accounts is an administrator instead of itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Hugging Face models, Perplexity\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: ChatGPT, Hugging Face models, Perplexity. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock is non-compliant\", \"finding\": \"Found: Auto-Lock Never. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block\nREQUIRED: block\nDETAIL: Not configured: Windows default (Block) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Compliant\", \"title\": \"Public inbound is compliant\", \"finding\": \"Public inbound is set to block, in line with the required block.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Found: partial. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: Claude Code. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: Veeam Agent for Microsoft Windows.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Found: date not readable. Confirm the last backup date in: Veeam Agent for Microsoft Windows.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically is non-compliant\", \"finding\": \"Found: set by hand. Required: set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › General › Date & Time: turn on Set Automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain inbound: does not meet the baseline\", \"finding\": \"Domain inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Tamper protection\", \"finding\": \"Found: off. Required: on. Tamper protection stops malware or users switching off Defender.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Attackers commonly disable antivirus before deploying ransomware.\"}, \"recommendation\": [\"Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Tamper protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Compliant\", \"title\": \"Account lockout duration is compliant\", \"finding\": \"Found: 30 minutes. This satisfies the baseline requirement of >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"Found: none. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised is non-compliant\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAI TOOLS FOUND: Hugging Face models, Ollama, Claude Code\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: C:\\Users\\jdoe\\.ollama\\models (59 GB). Local models: C:\\Users\\jdoe\\.cache\\huggingface\\hub (21 GB).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Hugging Face models, Ollama, Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: download only\nREQUIRED: download and install on\nDETAIL: Updates wait for the user to install them.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic iOS updates\", \"finding\": \"Automatic iOS updates is download only instead of download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Breach response procedure is no instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: auditd running, journal volatile\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging: meets the baseline\", \"finding\": \"The device reports auditd running, journal volatile, which meets the requirement (auditd running).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Operating system AI features restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: none built in\nREQUIRED: restricted by policy\nDETAIL: Linux distributions ship no built-in AI assistant.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (linux):\n- No action needed: Linux distributions have no built-in AI assistant.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Operating system AI features restricted: not assessed on this device\", \"finding\": \"Not applicable on this device: Linux distributions ship no built-in AI assistant.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (lmuller)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Signed-in user account type is Administrator (lmuller) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Found: partial. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (linux):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not applicable\", \"finding\": \"This check does not apply: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: download and install on\nREQUIRED: download and install on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic iOS updates is compliant\", \"finding\": \"Found: download and install on. This satisfies the baseline requirement of download and install on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: XProtect only\nREQUIRED: on-access scanning\nDETAIL: XProtect scans at launch and download; macOS has no built-in on-access scanning.\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (macos):\n- Not provided by macOS itself; deploy a managed endpoint protection product if required.\nverify: Re-run the audit; On-access malware protection should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"NotApplicable\", \"title\": \"On-access malware protection: not applicable\", \"finding\": \"On-access malware protection is not assessed here: XProtect scans at launch and download; macOS has no built-in on-access scanning.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: security patch 2026-10 (0 months old)\nREQUIRED: <= 90 days old\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Compliant\", \"title\": \"Android security update: meets the baseline\", \"finding\": \"Found: security patch 2026-10 (0 months old). This satisfies the baseline requirement of <= 90 days old. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Real-time protection [malware.realtime] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Real-time protection scans files and processes as they are used.\nrisk: High - Without it, malware is only found by scheduled scans, after it has run.\nfix (windows):\n- Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\nverify: Re-run the audit; Real-time protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.realtime\", \"status\": \"Compliant\", \"title\": \"Real-time protection: meets the baseline\", \"finding\": \"Real-time protection is set to on, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.realtime", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Where to find Automatic iOS updates on iPhone: Automatic iOS updates on iPhone: What do Automatic Updates show? Where to look: Settings › General › Software Update › Automatic Updates\n[2] Automatic iOS updates on iPhone: Automatic iOS updates on iPhone. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on iPhone: (1) Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files. How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] iOS security updates on iPhone: iOS security updates on iPhone. Why it matters: Security updates that are offered but not installed leave known, published vulnerabilities open. Risk if missing: High - Attackers target published vulnerabilities quickly after the fix is released. Requirement: Security updates offered by the OS vendor but not yet installed <= MAX_PENDING_SECURITY_UPDATES. Phones: iOS is at the latest security release known to the vulnerability data; Android security patch level is at most MOBILE_PATCH_MAX_AGE_DAYS old. How to fix on iPhone: (1) Settings › General › Software Update: install the latest iOS version offered. How to verify: Re-run the audit; pending security updates should be 0 or fewer. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience); GDPR Art. 32(1)(d) (Security of processing: regular testing and evaluation).\n[4] Supported operating system version on iPhone: Supported operating system version on iPhone. Why it matters: An operating system that no longer receives security updates cannot be kept secure. Risk if missing: High - New vulnerabilities in an unsupported OS are never fixed. Requirement: The OS version still receives security updates (minimums in the baseline). Phones: IOS_MIN_MAJOR / ANDROID_MIN_MAJOR. How to fix on iPhone: (1) Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone. How to verify: Re-run the audit; Supported operating system version should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nQUESTION: How do I fix ios security updates on iPhone?"}, {"role": "assistant", "content": "Security updates that are offered but not installed leave known, published vulnerabilities open. To fix it on iPhone [3]:\n1. Settings › General › Software Update: install the latest iOS version offered.\n\nTo check it worked: Re-run the audit; pending security updates should be 0 or fewer."}], "meta": {"task": "chat", "os": "ios", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically: does not meet the baseline\", \"finding\": \"Date & time set automatically is set by hand instead of set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › General › Date & Time: turn on Set Automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Hugging Face models\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Hugging Face models.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Real-time protection [malware.realtime] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Real-time protection scans files and processes as they are used.\nrisk: High - Without it, malware is only found by scheduled scans, after it has run.\nfix (windows):\n- Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\nverify: Re-run the audit; Real-time protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.realtime\", \"status\": \"Compliant\", \"title\": \"Real-time protection matches the required setting\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.realtime", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: automatic app updates on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Pending\", \"title\": \"App updates is pending\", \"finding\": \"The user could not find or confirm this setting. Current state: not answered. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound is non-compliant\", \"finding\": \"Private inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Standard (akaya)\nREQUIRED: Standard\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Compliant\", \"title\": \"Signed-in user account type: meets the baseline\", \"finding\": \"The device reports Standard (akaya), which meets the requirement (Standard).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iPadOS 26.5 (ios)\nAI TOOLS FOUND: DeepSeek, Gemini, Replika\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- DeepSeek: Cloud AI assistant; provider DeepSeek (China); data: Provider cloud in the People's Republic of China; default risk minimal; GDPR: No EU adequacy decision for China and no business data processing terms; EU data protection authorities have restricted or investigated it. Do not use with work or personal data.\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\n- Replika: AI companion chatbot; provider Luka (US); data: Provider cloud outside the EEA; default risk limited; GDPR: Consumer companion app that collects sensitive personal conversations and has been subject to EU data protection enforcement; not suitable for work use.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"DeepSeek\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No EU adequacy decision for China and no business data processing terms; EU data protection authorities have restricted or investigated it. Do not use with work or personal data.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Replika\", \"kind\": \"AI companion chatbot\", \"approved\": false, \"default_risk_class\": \"limited\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer companion app that collects sensitive personal conversations and has been subject to EU data protection enforcement; not suitable for work use.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: DeepSeek, Gemini, Replika.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 15.7.2 (macos)\nHOSTNAME: OPS-MacBook-03\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=30, Non-Compliant=19, NotApplicable=3, Pending=3, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Error | not read\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5246\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5250, 16 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode on\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 1200 idle, password delay 5\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (jdoe)\n- encryption.enabled | Encryption | Non-Compliant | FileVault off\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no FileVault\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 42 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 225 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 15.7.2\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Perplexity, ChatGPT, Claude\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Perplexity, ChatGPT, Claude\n- aiact.risk_classification | AI system risk classification | Pending | Perplexity, ChatGPT, Claude\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- lockout.observation: Low\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.processor_agreements: High\n- org.ai_register: Medium\nSCORE (computed): compliant=30, assessed=49, percent=61\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode on\n- encryption.enabled | Encryption | High | FileVault off\n- encryption.protection | Disk encryption protection | High | not encrypted\n- vuln.pending_updates | Pending security updates | High | 2 pending\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- encryption.enabled: Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM."}, {"role": "assistant", "content": "{\"device\": \"OPS-MacBook-03 - macOS 15.7.2\", \"score\": {\"compliant\": 30, \"assessed\": 49, \"percent\": 61}, \"counts\": {\"Compliant\": 30, \"Non-Compliant\": 19, \"NotApplicable\": 3, \"Pending\": 3, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode on\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"FileVault off\"}, {\"check_id\": \"encryption.protection\", \"name\": \"Disk encryption protection\", \"risk\": \"High\", \"found\": \"not encrypted\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"2 pending\"}], \"summary\": \"OPS-MacBook-03 meets 30 of 49 assessed checks (61%). The most serious gaps are Firewall, Firewall inbound policy and Encryption. 1 checks could not be read. 3 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Encryption: Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (windows):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not applicable\", \"finding\": \"Personal data discovery is not assessed here: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Google backup is non-compliant\", \"finding\": \"Google backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: usage reporting off\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Error\", \"title\": \"Diagnostic data minimised: could not be checked\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public outbound\", \"finding\": \"Found: block. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates: does not meet the baseline\", \"finding\": \"Found: automatic app updates off. Required: automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iPadOS 27.0 (ios)\nHOSTNAME: OPS-iPad-28\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=12, Non-Compliant=4, NotApplicable=3, Pending=10, Error=0\nEVIDENCE: 2 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Pending | not answered\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iPadOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | off\n- gdpr.os_supported | Supported operating system version | Compliant | iPadOS 27.0\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Pending | not answered\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.timeout: Medium\n- mobile.app_sources: High\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\nSCORE (computed): compliant=12, assessed=16, percent=75\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- gdpr.auto_updates | Automatic iOS updates | High | off\n- gdpr.backup_configured | iCloud Backup | High | no backup\n- screen_lock.timeout | Auto-Lock | Medium | Auto-Lock Never\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- gdpr.auto_updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\n- gdpr.backup_configured: Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps."}, {"role": "assistant", "content": "{\"device\": \"OPS-iPad-28 - iPadOS 27.0\", \"score\": {\"compliant\": 12, \"assessed\": 16, \"percent\": 75}, \"counts\": {\"Compliant\": 12, \"Non-Compliant\": 4, \"NotApplicable\": 3, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Auto-Lock\", \"risk\": \"Medium\", \"found\": \"Auto-Lock Never\"}], \"summary\": \"OPS-iPad-28 meets 12 of 16 assessed checks (75%). The most serious gaps are Apps only from the App Store, Automatic iOS updates and iCloud Backup. 10 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Automatic iOS updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\", \"iCloud Backup: Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iPadOS 18.7.9 (ios)\nHOSTNAME: HR-iPad-57\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=5, Non-Compliant=20, NotApplicable=0, Pending=4, Error=0\nEVIDENCE: 0 detected by the browser, 21 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Non-Compliant | no passcode\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Non-Compliant | off (no passcode)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iPadOS 18.7.9 (update to 18.7.10 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 160 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | download only\n- gdpr.os_supported | Supported operating system version | Compliant | iPadOS 18.7.9\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | Gemini, Grok\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Gemini, Grok\n- aiact.risk_classification | AI system risk classification | Pending | Gemini, Grok\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | ChatGPT extension off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- encryption.enabled: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=5, assessed=25, percent=20\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- encryption.enabled | Data protection (encryption) | High | off (no passcode)\n- vuln.pending_updates | iOS security updates | High | iPadOS 18.7.9 (update to 18.7.10 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 160 critical/high\n- gdpr.auto_updates | Automatic iOS updates | High | download only\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- encryption.enabled: Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered."}, {"role": "assistant", "content": "{\"device\": \"HR-iPad-57 - iPadOS 18.7.9\", \"score\": {\"compliant\": 5, \"assessed\": 25, \"percent\": 20}, \"counts\": {\"Compliant\": 5, \"Non-Compliant\": 20, \"NotApplicable\": 0, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Data protection (encryption)\", \"risk\": \"High\", \"found\": \"off (no passcode)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iPadOS 18.7.9 (update to 18.7.10 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"160 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"download only\"}], \"summary\": \"HR-iPad-57 meets 5 of 25 assessed checks (20%). The most serious gaps are Apps only from the App Store, Data protection (encryption) and iOS security updates. 4 items are pending. 21 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Data protection (encryption): Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: none; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging: does not meet the baseline\", \"finding\": \"Found: logon: none; account management: none. Required: logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: itadmin\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Compliant\", \"title\": \"Approved IT administrator: meets the baseline\", \"finding\": \"Found: itadmin. This satisfies the baseline requirement of itadmin.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: passcode set\nREQUIRED: passcode set\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Passcode is compliant\", \"finding\": \"Passcode is set to passcode set, in line with the required passcode set. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Disk encryption recovery on macOS: Disk encryption recovery on macOS. Why it matters: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes. Risk if missing: Medium - Without a recovery key, an encrypted disk can become permanently unreadable. Requirement: A recovery key or second unlock method exists. How to fix on macOS: (1) Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM. How to verify: Re-run the audit; Disk encryption recovery should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.13 (Information backup); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption); GDPR Art. 32(1)(c) (Security of processing: restore availability and access after an incident).\n[2] Disk encryption protection on macOS: Disk encryption protection on macOS. Why it matters: Encryption only protects the data when it is active, fully applied and uses a strong cipher. Risk if missing: High - Suspended or partial encryption leaves data readable on a lost device. Requirement: Encryption protection is active (not suspended or in progress) with a strong cipher. How to fix on macOS: (1) Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress. How to verify: Re-run the audit; Disk encryption protection should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[3] Encryption on macOS: Encryption on macOS. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: The system disk is fully encrypted (BitLocker / LUKS / FileVault). Phones: iOS data protection is on when a passcode is set; Android 10+ encrypts by default. How to fix on macOS: (1) Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[4] ISO 27001:2022 A.8.24 Use of cryptography: ISO 27001:2022 A.8.24 (Use of cryptography): Define and apply rules for using cryptography, including key management. An auditor looks for: Full-disk encryption with strong algorithms; Recovery keys stored safely. On endpoints: Disk encryption protects data on lost or stolen devices.\n\nMY AUDIT: macOS 26.5.2 (macos), audited 2026-10-05; 30 of 50 assessed checks compliant; 20 gaps, 3 pending, 0 not read. Gaps: Minimum password length: 12; Password expiration: never; Password complexity: not enforced; Apple malware protection service: XProtect scanning on; Gatekeeper assessments disabled; Firewall: off; Firewall inbound policy: firewall off, stealth mode off; USB storage write: allow; Disk encryption recovery: no recovery key.\n\nQUESTION: How do I fix disk encryption recovery?"}, {"role": "assistant", "content": "Your audit found Disk encryption recovery: no recovery key. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes. To fix it [1]:\n1. Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "macos", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 26.7 (ios)\nAI TOOLS FOUND: Character.AI\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Character.AI: AI companion chatbot; provider Character Technologies (US); data: Provider cloud outside the EEA; default risk limited; GDPR: Consumer entertainment chatbot; not suitable for work use. Users must be told they are talking to an AI (AI Act Art. 50).\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Character.AI\", \"kind\": \"AI companion chatbot\", \"approved\": false, \"default_risk_class\": \"limited\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer entertainment chatbot; not suitable for work use. Users must be told they are talking to an AI (AI Act Art. 50).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Character.AI.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apple Intelligence external integrations restricted\", \"finding\": \"The device reports not restricted, but the baseline requires restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised is non-compliant\", \"finding\": \"Usage & diagnostics sharing minimised is usage & diagnostics on instead of usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Found: 10. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: lmuller\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"Found: lmuller. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: type NoSync, service Stopped/Manual\nREQUIRED: NTP or domain time\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (windows):\n- w32tm /config /syncfromflags:manual /manualpeerlist:\"pool.ntp.org\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"The device reports type NoSync, service Stopped/Manual, but the baseline requires NTP or domain time. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"w32tm /config /syncfromflags:manual /manualpeerlist:\\\"pool.ntp.org\\\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAI TOOLS FOUND: ChatGPT, Hugging Face models\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: C:\\Users\\jdoe\\.cache\\huggingface\\hub (10 GB).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Hugging Face models.\"}"}], "meta": {"task": "ai_classification", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iPadOS 18.7.8 (ios)\nHOSTNAME: FIN-iPad-22\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=13, Non-Compliant=6, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 2 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 1 minute\n- mobile.passcode_strength | Passcode strength | Compliant | 6-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iPadOS 18.7.8 (update to 18.7.10 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 189 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iPadOS 18.7.8\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | Gemini\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Gemini\n- aiact.risk_classification | AI system risk classification | Pending | Gemini\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | ChatGPT extension off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.app_sources: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- aiact.unapproved: Medium\nSCORE (computed): compliant=13, assessed=19, percent=68\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- vuln.pending_updates | iOS security updates | High | iPadOS 18.7.8 (update to 18.7.10 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 189 critical/high\n- gdpr.backup_configured | iCloud Backup | High | no backup\n- vuln.outdated_apps | App updates | Medium | automatic app updates off\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"FIN-iPad-22 - iPadOS 18.7.8\", \"score\": {\"compliant\": 13, \"assessed\": 19, \"percent\": 68}, \"counts\": {\"Compliant\": 13, \"Non-Compliant\": 6, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iPadOS 18.7.8 (update to 18.7.10 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"189 critical/high\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"vuln.outdated_apps\", \"name\": \"App updates\", \"risk\": \"Medium\", \"found\": \"automatic app updates off\"}], \"summary\": \"FIN-iPad-22 meets 13 of 19 assessed checks (68%). The most serious gaps are Apps only from the App Store, iOS security updates and Known vulnerabilities in iOS. 10 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: NTP service yes, synchronised yes\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation: meets the baseline\", \"finding\": \"Time synchronisation is set to NTP service yes, synchronised yes, in line with the required time synchronised.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2026-06 (4 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2026-06 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Android security update is non-compliant\", \"finding\": \"Found: security patch 2026-06 (4 months old). Required: <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification: not assessed on this device\", \"finding\": \"Not applicable on this device: no AI tools found. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private firewall [firewall.private.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Private firewall is non-compliant\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Perplexity\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: ChatGPT, Perplexity. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: DEVSEIS\\itadmin\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Compliant\", \"title\": \"Approved IT administrator is compliant\", \"finding\": \"Found: DEVSEIS\\\\itadmin. This satisfies the baseline requirement of itadmin.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Data processing agreements is partial instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 12 (android)\nHOSTNAME: legal-pixel-36\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=10, Non-Compliant=10, NotApplicable=2, Pending=8, Error=0\nEVIDENCE: 1 detected by the browser, 21 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Non-Compliant | Screen timeout 10 minutes\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | pattern\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 12 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2026-06 (4 months old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 12\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- gdpr.os_supported: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=10, assessed=20, percent=50\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- vuln.pending_updates | Android security update | High | security patch 2026-06 (4 months old)\n- gdpr.os_supported | Supported operating system version | High | Android 12\n- gdpr.backup_configured | Google backup | High | no backup\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone."}, {"role": "assistant", "content": "{\"device\": \"legal-pixel-36 - Android 12\", \"score\": {\"compliant\": 10, \"assessed\": 20, \"percent\": 50}, \"counts\": {\"Compliant\": 10, \"Non-Compliant\": 10, \"NotApplicable\": 2, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2026-06 (4 months old)\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 12\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Google backup\", \"risk\": \"High\", \"found\": \"no backup\"}], \"summary\": \"legal-pixel-36 meets 10 of 20 assessed checks (50%). The most serious gaps are Google Play Protect, Install unknown apps blocked and Android security update. 8 items are pending. 21 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 70 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Found: 70 days ago. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound: does not meet the baseline\", \"finding\": \"Private inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Google Play Protect is non-compliant\", \"finding\": \"Google Play Protect is Play Protect scanning off instead of on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training matches the required setting\", \"finding\": \"AI literacy training is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off (no passcode)\nREQUIRED: on\nDETAIL: Without a passcode, iOS data protection is not active.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection (encryption)\", \"finding\": \"Data protection (encryption) is off (no passcode) instead of on. Full-disk encryption keeps data unreadable if the device is lost or stolen. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Device encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (Android 14 encrypts by default)\nREQUIRED: on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (android):\n- Update to a supported Android version (10 or later encrypts by default) or replace the phone; set a PIN or password screen lock.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Device encryption is compliant\", \"finding\": \"Found: on (Android 14 encrypts by default). This satisfies the baseline requirement of on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: all apps: Not allowed\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer).\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Pending\", \"title\": \"Install unknown apps blocked: decision pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Screen lock password requirement on Linux: Screen lock password requirement on Linux. Why it matters: Unlocking the screen must require the user's password. Risk if missing: Medium - A screen lock that needs no password does not protect anything. Requirement: A password is required to unlock (macOS: within 5 seconds of the screen locking). How to fix on Linux: (1) Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false. How to verify: Re-run the audit; the password requirement should be Compliant. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Screen lock timeout on Linux: Screen lock timeout on Linux. Why it matters: The shorter the idle time before locking, the smaller the window for misuse. Risk if missing: Medium - A long or missing timeout leaves unattended devices open. Requirement: Idle time before lock (including any grace delay) is 1..SCREEN_LOCK_MAX_SECONDS seconds (baseline: 300 = 5 minutes). Phones: Auto-Lock / screen timeout. How to fix on Linux: (1) gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings. How to verify: Re-run the audit; Screen lock timeout should be 300 seconds or less. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Operating system AI features restricted on Linux: Operating system AI features restricted on Linux. Why it matters: Built-in OS AI features can capture screen content or send text to external AI services. Risk if missing: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose. Requirement: OS-level AI features that capture or send user data are disabled by policy. NotApplicable where the OS has none. Phones: external AI integrations (ChatGPT in Apple Intelligence, assistant access to screen content) are off unless approved. How to fix on Linux: (1) No action needed: Linux distributions have no built-in AI assistant. How to verify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[4] Screen lock on Linux: Screen lock on Linux. Why it matters: An idle device should lock itself so nobody can use it while the owner is away. Risk if missing: Medium - An unlocked, unattended device exposes all the user's data and accounts. Requirement: The screen locks when idle (Windows: machine inactivity limit or a password-protected screen saver). Phones: a passcode / PIN / password screen lock is set. How to fix on Linux: (1) For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/. How to verify: Re-run the audit; Screen lock should be on. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer (x86-64), Chrome 141; detected by the browser, security settings not checked.\n\nQUESTION: How do I fix screen lock on Linux?"}, {"role": "assistant", "content": "An idle device should lock itself so nobody can use it while the owner is away. To fix it on Linux [4]:\n1. For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\n\nTo check it worked: Re-run the audit; Screen lock should be on."}], "meta": {"task": "chat", "os": "linux", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"Human oversight for high-risk AI is partial instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted: does not meet the baseline\", \"finding\": \"Found: not restricted. Required: restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"The device reports anydesk, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: guest login off\nREQUIRED: guest login off\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (macos):\n- sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled is compliant\", \"finding\": \"Found: guest login off. This satisfies the baseline requirement of guest login off.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: LM Studio, Claude Code, Hugging Face models\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: LM Studio, Claude Code, Hugging Face models. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Found: USB drives writable without encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: no guest login\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled matches the required setting\", \"finding\": \"Found: no guest login. This satisfies the baseline requirement of no guest login.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: type NoSync, service Stopped/Manual\nREQUIRED: NTP or domain time\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (windows):\n- w32tm /config /syncfromflags:manual /manualpeerlist:\"pool.ntp.org\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"The device reports type NoSync, service Stopped/Manual, but the baseline requires NTP or domain time. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"w32tm /config /syncfromflags:manual /manualpeerlist:\\\"pool.ntp.org\\\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"The device reports no, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: AllowTelemetry=3 (Settings)\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Found: AllowTelemetry=3 (Settings). Required: required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"Account lockout duration is no lockout instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 14\nDETAIL: pwpolicy global account policy\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (macos):\n- Deploy a Passcode configuration profile through your MDM with minLength = 14.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Minimum password length is not set instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile through your MDM with minLength = 14.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: only: none\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Error\", \"title\": \"Unapproved AI tools (shadow AI): evidence missing\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: dnf-automatic timer on, apply_updates=yes\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates: meets the baseline\", \"finding\": \"Automatic security updates is set to dnf-automatic timer on, apply_updates=yes, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: none; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging is non-compliant\", \"finding\": \"Found: logon: none; account management: none. Required: logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nHOSTNAME: DESKTOP-UNG1VHJ\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=37, Non-Compliant=12, NotApplicable=9, Pending=6, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 12\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | Sophos Endpoint (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; too long or off for: FABRIKAM\\asmith (600s, active=0)\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | FABRIKAM\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-UNG1VHJ\\asmith\n- user.account_type | Signed-in user account type | Compliant | Standard (FABRIKAM\\asmith)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 221 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 22 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 22631\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | File History (jdoe)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=1 (policy)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.expiration: Low\n- password.complexity: Medium\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.removable_encryption: Medium\n- org.ropa: High\n- org.processor_agreements: High\nSCORE (computed): compliant=37, assessed=49, percent=76\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | DESKTOP-UNG1VHJ\\asmith\n- vuln.pending_updates | Pending security updates | High | 2 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 221 critical/high\n- org.ropa | Record of processing activities | High | partial\n- org.processor_agreements | Data processing agreements | High | no\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\n- vuln.pending_updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-UNG1VHJ - Microsoft Windows 11 Pro 10.0.22631\", \"score\": {\"compliant\": 37, \"assessed\": 49, \"percent\": 76}, \"counts\": {\"Compliant\": 37, \"Non-Compliant\": 12, \"NotApplicable\": 9, \"Pending\": 6, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"DESKTOP-UNG1VHJ\\\\asmith\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"2 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"221 critical/high\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"org.processor_agreements\", \"name\": \"Data processing agreements\", \"risk\": \"High\", \"found\": \"no\"}], \"summary\": \"DESKTOP-UNG1VHJ meets 37 of 49 assessed checks (76%). The most serious gaps are Unauthorized employee local administrator accounts, Pending security updates and Known vulnerabilities in installed software. 6 items are pending.\", \"next_steps\": [\"Unauthorized employee local administrator accounts: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\", \"Pending security updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 5 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: OneDrive folder backup (policy).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Confirm the last backup date in: OneDrive folder backup (policy). Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Found: none. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device is non-compliant\", \"finding\": \"The user reports off, but the baseline requires on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Compliant\", \"title\": \"Firewall outbound policy: meets the baseline\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no signature database\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures is non-compliant\", \"finding\": \"Found: no signature database. Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 10 Pro build 19045\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"The device reports Microsoft Windows 10 Pro build 19045, but the baseline requires build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Domain outbound\", \"finding\": \"Found: block. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI assistant access restricted\", \"finding\": \"AI assistant access restricted is assistant screen access on instead of screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Answer ORG_DPIA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 202 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-03-16  06:45:48\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"The device reports 202 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates is non-compliant\", \"finding\": \"Found: freshclam not running. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: File History (jdoe).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Found: date not readable. Confirm the last backup date in: File History (jdoe).\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training matches the required setting\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Restrict external storage with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read: meets the baseline\", \"finding\": \"USB storage read is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator is non-compliant\", \"finding\": \"Approved IT administrator is none of the approved accounts is an administrator instead of itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an enabled administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"The device reports none of the approved accounts is an enabled administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ewagner, pmartin\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee administrator accounts\", \"finding\": \"Unauthorized employee administrator accounts is ewagner, pmartin instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"Found: none. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection is non-compliant\", \"finding\": \"Found: not running. Required: on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Microsoft Defender Antimalware [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: service True, antivirus False\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (windows):\n- Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Microsoft Defender Antimalware\", \"finding\": \"The device reports service True, antivirus False, but the baseline requires running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: logon: success+failure; account management: success+failure\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging: meets the baseline\", \"finding\": \"The device reports logon: success+failure; account management: success+failure, which meets the requirement (logon success+failure and account changes audited).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"Diagnostic data minimised is not configured instead of required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 14\nDETAIL: pwpolicy global account policy\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (macos):\n- Deploy a Passcode configuration profile through your MDM with minLength = 14.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"The device reports 12, but the baseline requires >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile through your MDM with minLength = 14.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.5.2 (macos)\nHOSTNAME: MKT-MacBook-44\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=25, Non-Compliant=23, NotApplicable=4, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout\n- malware.antivirus | Antivirus | Compliant | XProtect 5321\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5215, 28 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode on\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 180 idle, password delay 300\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after 60 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | rschmit\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (mkhan)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no recovery key\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 41 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 228 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.5.2\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude, Cursor, Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude, Cursor, Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Claude, Cursor, Claude Code\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=25, assessed=48, percent=52\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode on\n- admin.unauthorized | Unauthorized employee administrator accounts | High | rschmit\n- user.account_type | Signed-in user account type | High | Administrator (mkhan)\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 41 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin."}, {"role": "assistant", "content": "{\"device\": \"MKT-MacBook-44 - macOS 26.5.2\", \"score\": {\"compliant\": 25, \"assessed\": 48, \"percent\": 52}, \"counts\": {\"Compliant\": 25, \"Non-Compliant\": 23, \"NotApplicable\": 4, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode on\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"rschmit\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (mkhan)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"41 critical/high\"}], \"summary\": \"MKT-MacBook-44 meets 25 of 48 assessed checks (52%). The most serious gaps are Firewall, Firewall inbound policy and Unauthorized employee administrator accounts. 4 items are pending.\", \"next_steps\": [\"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.6 (ios)\nHOSTNAME: SALES-iPhone-28\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=17, Non-Compliant=8, NotApplicable=3, Pending=1, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store and employer-managed apps\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.6 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 42 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.6\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | backed up to a computer\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- org.processor_agreements: High\n- org.ai_transparency: Low\nSCORE (computed): compliant=17, assessed=25, percent=68\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | iOS security updates | High | iOS 26.6 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 42 critical/high\n- org.processor_agreements | Data processing agreements | High | no\n- screen_lock.timeout | Auto-Lock | Medium | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Medium | 4-digit code\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- org.processor_agreements: List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers."}, {"role": "assistant", "content": "{\"device\": \"SALES-iPhone-28 - iOS 26.6\", \"score\": {\"compliant\": 17, \"assessed\": 25, \"percent\": 68}, \"counts\": {\"Compliant\": 17, \"Non-Compliant\": 8, \"NotApplicable\": 3, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.6 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"42 critical/high\"}, {\"check_id\": \"org.processor_agreements\", \"name\": \"Data processing agreements\", \"risk\": \"High\", \"found\": \"no\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Auto-Lock\", \"risk\": \"Medium\", \"found\": \"Auto-Lock Never\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Passcode strength\", \"risk\": \"Medium\", \"found\": \"4-digit code\"}], \"summary\": \"SALES-iPhone-28 meets 17 of 25 assessed checks (68%). The most serious gaps are iOS security updates, Known vulnerabilities in iOS and Data processing agreements. 1 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Data processing agreements: List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall outbound policy is non-compliant\", \"finding\": \"The device reports no firewall, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: sali, asmith\nREQUIRED: password required\nDETAIL: Users checked: sali, asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"Found: not required for: sali, asmith. Required: password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: ChatGPT.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"Account lockout duration is 10 minutes instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 20\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"Account lockout threshold is 20 instead of 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Found: USB storage writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pattern\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: Pattern locks are easy to guess from screen smudges or by watching.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength is non-compliant\", \"finding\": \"The user reports pattern, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Microsoft Copilot\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: ChatGPT, Microsoft Copilot. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\", \"For a stand-alone device: net accounts /maxpwage:365\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (macos):\n- Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator is non-compliant\", \"finding\": \"Approved IT administrator is none of the approved accounts is an administrator instead of itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 416 source packages checked against ubuntu:24.04 advisories. 0 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Compliant\", \"title\": \"Known vulnerabilities in installed software: meets the baseline\", \"finding\": \"The device reports 0 critical/high, which meets the requirement (<= 0 critical/high).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain inbound is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Screen timeout 10 minutes\nREQUIRED: <= 300 seconds\nDETAIL: The phone stays unlocked for 10 minutes when not in use.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen timeout\", \"finding\": \"The user reports Screen timeout 10 minutes, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password length\", \"finding\": \"The device reports 0, but the baseline requires >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: root filesystem not encrypted\nREQUIRED: system disk encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (linux):\n- Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"The device reports root filesystem not encrypted, but the baseline requires system disk encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"Account lockout duration is no lockout configured instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 180 seconds\nREQUIRED: <= 300 seconds\nDETAIL: Idle 180s (screen saver/display sleep) + 0s password delay.\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout matches the required setting\", \"finding\": \"Screen lock timeout is set to 180 seconds, in line with the required <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 16 (android)\nHOSTNAME: dev-galaxy-44\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=15, Non-Compliant=7, NotApplicable=0, Pending=8, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 1 minute\n- mobile.passcode_strength | Screen lock strength | Compliant | 6-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Pending | not answered\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 16 encrypts by default)\n- vuln.pending_updates | Android security update | Pending | not answered\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Compliant | Android 16\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Grammarly, Microsoft Copilot\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Grammarly, Microsoft Copilot\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Grammarly, Microsoft Copilot\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=15, assessed=22, percent=68\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- org.breach_procedure | Breach response procedure | High | no\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | ChatGPT, Grammarly, Microsoft Copilot\n- org.dpia | Data protection impact assessment | Medium | partial\n- org.human_oversight | Human oversight for high-risk AI | Medium | partial\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- org.breach_procedure: Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it."}, {"role": "assistant", "content": "{\"device\": \"dev-galaxy-44 - Android 16\", \"score\": {\"compliant\": 15, \"assessed\": 22, \"percent\": 68}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 7, \"NotApplicable\": 0, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"org.breach_procedure\", \"name\": \"Breach response procedure\", \"risk\": \"High\", \"found\": \"no\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"ChatGPT, Grammarly, Microsoft Copilot\"}, {\"check_id\": \"org.dpia\", \"name\": \"Data protection impact assessment\", \"risk\": \"Medium\", \"found\": \"partial\"}, {\"check_id\": \"org.human_oversight\", \"name\": \"Human oversight for high-risk AI\", \"risk\": \"Medium\", \"found\": \"partial\"}], \"summary\": \"dev-galaxy-44 meets 15 of 22 assessed checks (68%). The most serious gaps are Google Play Protect and Breach response procedure. 8 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Breach response procedure: Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Answer the 4 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 207 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 18.7 checked against NVD iOS records. iOS 18.7 (207 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2025-14174, CVE-2025-31255). 356 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in iOS\", \"finding\": \"The device reports 207 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device is non-compliant\", \"finding\": \"The user reports off, but the baseline requires on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Compliant\", \"title\": \"Human oversight for high-risk AI is compliant\", \"finding\": \"Human oversight for high-risk AI is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 3 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Tahoe 26.2.2-24H202, Command Line Tools for Xcode-26.0, Safari18.6Tahoe-26.1.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"The device reports 3 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Compliant\", \"title\": \"AI register matches the required setting\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: USB mass storage driver disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Non-Compliant\", \"title\": \"USB storage device access is non-compliant\", \"finding\": \"USB storage device access is deny instead of allow. Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The device access setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage device access should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Found: not set. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with pinHistory = 24.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption protection\", \"finding\": \"Found: not encrypted. Required: fully encrypted. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 19 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: systemd, openssl, libsystemd0. Package lists are 30 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"The device reports 19 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 45 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 45\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"Found: 45 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Removable media encryption is external drives writable without enforced encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: guest login off\nREQUIRED: guest login off\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (macos):\n- sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled is compliant\", \"finding\": \"The device reports guest login off, which meets the requirement (guest login off).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no passcode\nREQUIRED: passcode set\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Passcode is non-compliant\", \"finding\": \"Found: no passcode. Required: passcode set. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker\", \"finding\": \"The device reports off, but the baseline requires on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 14.7.8 (macos)\nAI TOOLS FOUND: Ollama, Claude\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local AI service listening on port 11434. Local models: /Users/jdoe/.ollama/models (59G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Ollama, Claude.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: passcode set\nREQUIRED: passcode set\nDETAIL: The browser reports that a device unlock method (passcode, PIN, Face ID or fingerprint) is set up.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Passcode: meets the baseline\", \"finding\": \"Found: passcode set. This satisfies the baseline requirement of passcode set.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Current state: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Unapproved AI tools (shadow AI) on Android: Unapproved AI tools (shadow AI) on Android. Why it matters: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Risk if missing: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations. Requirement: Every AI tool found is listed in APPROVED_AI_TOOLS. How to fix on Android: (1) Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it. (2) Publish an AI acceptable-use policy so staff know which tools are allowed. How to verify: Re-run the audit; Unapproved AI tools should read none. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 28 (Processor); GDPR Art. 44 (General principle for transfers); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[2] Password expiration on macOS: Password expiration on macOS. Why it matters: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful. Risk if missing: Low - A leaked password can stay valid indefinitely. Requirement: Passwords expire after 1..PASSWORD_MAX_AGE_DAYS days (0 in the baseline = expiry not required). How to fix on macOS: (1) Deploy a Passcode configuration profile with maxPINAgeInDays = 365. How to verify: Re-run the audit; Password expiration should be between 1 and 365 days. Related controls: ISO 27001:2022 A.5.17 (Authentication information); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Password expiration on Linux: Password expiration on Linux. Why it matters: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful. Risk if missing: Low - A leaked password can stay valid indefinitely. Requirement: Passwords expire after 1..PASSWORD_MAX_AGE_DAYS days (0 in the baseline = expiry not required). How to fix on Linux: (1) Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts). (2) Apply it to existing accounts: chage --maxdays 365 <user> How to verify: Re-run the audit; Password expiration should be between 1 and 365 days. Related controls: ISO 27001:2022 A.5.17 (Authentication information); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Password expiration on Windows: Password expiration on Windows. Why it matters: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful. Risk if missing: Low - A leaked password can stay valid indefinitely. Requirement: Passwords expire after 1..PASSWORD_MAX_AGE_DAYS days (0 in the baseline = expiry not required). How to fix on Windows: (1) Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days. (2) For a stand-alone device: net accounts /maxpwage:365 How to verify: Re-run the audit; Password expiration should be between 1 and 365 days. Related controls: ISO 27001:2022 A.5.17 (Authentication information); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Android 15 (android) on a Pixel 8, Chrome 141; detected by the browser, security settings not checked.\n\nQUESTION: How do I fix a leaking tap?"}, {"role": "assistant", "content": "I can only help with device security, ISO 27001, GDPR and the EU AI Act, so I can't answer that. Ask me, for example, how to turn on disk encryption, whether an AI tool is OK for work, or run an audit of this device."}], "meta": {"task": "chat", "os": "any", "status": "oos", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.7 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates is non-compliant\", \"finding\": \"Found: iOS 26.7 (update to 26.7.1 available). Required: latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Compliant\", \"title\": \"Find My Device matches the required setting\", \"finding\": \"Find My Device is set to on, in line with the required on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: remotemanagement\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Found: remotemanagement. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: asmith\nREQUIRED: password required\nDETAIL: Users checked: asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"Found: not required for: asmith. Required: password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Character.AI\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: Character.AI. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"Found: partial. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength is non-compliant\", \"finding\": \"Passcode strength is 4-digit code instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: ClamAV (clamav-daemon)\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus matches the required setting\", \"finding\": \"The device reports ClamAV (clamav-daemon), which meets the requirement (antivirus running).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: FileVault off\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"Found: FileVault off. Required: FileVault on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised: does not meet the baseline\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"Firewall inbound policy is allow (no firewall) instead of deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 15 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Compliant\", \"title\": \"Account lockout duration is compliant\", \"finding\": \"The device reports 15 minutes, which meets the requirement (>= 15 minutes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: NTP service no, synchronised no\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Time synchronisation\", \"finding\": \"The device reports NTP service no, synchronised no, but the baseline requires time synchronised. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"timedatectl set-ntp true (or enable chronyd).\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Microsoft Defender Antimalware [malware.service] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: Defender in Passive Mode\nREQUIRED: running\nDETAIL: Another antivirus is primary: CrowdStrike Falcon\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (windows):\n- Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\nverify: Re-run the audit; the malware protection service should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"NotApplicable\", \"title\": \"Microsoft Defender Antimalware: not assessed on this device\", \"finding\": \"Microsoft Defender Antimalware is not assessed here: Another antivirus is primary: CrowdStrike Falcon.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Ollama\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is Ollama instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Compliant\", \"title\": \"Public firewall matches the required setting\", \"finding\": \"Public firewall is set to on, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Ollama, Claude\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Ollama, Claude.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery: does not meet the baseline\", \"finding\": \"BitLocker recovery is key protectors: Tpm instead of recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAI TOOLS FOUND: Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2026-05 (5 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2026-05 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Android security update\", \"finding\": \"The user reports security patch 2026-05 (5 months old), but the baseline requires <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.7 checked against NVD iOS records. iOS 26.7 (15 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-43688). 32 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS is non-compliant\", \"finding\": \"Found: 15 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Current state: Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Password history is 5 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\", \"For a stand-alone device: net accounts /uniquepw:24\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 12\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Chrome 141). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"Supported operating system version is Android 12 instead of >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"The device reports no, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery: does not meet the baseline\", \"finding\": \"BitLocker recovery is key protectors: Tpm instead of recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: XProtect 5276, 104 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Malware protection signatures\", \"finding\": \"Malware protection signatures is XProtect 5276, 104 days old instead of <= 30 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Unauthorized employee administrator accounts is mkhan instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall: does not meet the baseline\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Windows Defender (off, out of date)\nREQUIRED: enabled and up to date\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (windows):\n- Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Antivirus: does not meet the baseline\", \"finding\": \"Antivirus is Windows Defender (off, out of date) instead of enabled and up to date. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"The device reports no lockout, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: teamviewer\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Found: teamviewer. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 95 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 632 source packages checked against debian:13 advisories. apache-log4j2 2.15.0-1 (1 CVE, critical, actively exploited; fixed in 2.16.0-1; e.g. CVE-2021-45046); open-vm-tools 2:9.4.6-1770165-1 (7 CVEs, high, actively exploited; fixed in 2:12.5.0-2+deb13u1; e.g. CVE-2009-1143, CVE-2022-31676); qemu 1:3.1+dfsg-1 (40 CVEs, critical; fixed in 1:10.0.10+ds-0+deb13u1; e.g. CVE-2018-20125, CVE-2018-20191); suricata 1:7.0.6-1 (24 CVEs, critical; fixed in 1:7.0.10-1+deb13u4; e.g. CVE-2024-45795, CVE-2024-47187); nova 2013.2.2 (6 CVEs, critical; fixed in 2:31.0.0-6+deb13u2; e.g. CVE-2015-5162, CVE-2017-17051); and 12 more. 238 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 95 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 207 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 18.7 checked against NVD iOS records. iOS 18.7 (207 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2025-14174, CVE-2025-31255). 356 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS: does not meet the baseline\", \"finding\": \"The device reports 207 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Data protection impact assessment is no instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker is non-compliant\", \"finding\": \"BitLocker is off instead of on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"Found: no lockout. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Domain firewall: does not meet the baseline\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nHOSTNAME: LAPTOP-ZRCPWCK\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=14, Non-Compliant=38, NotApplicable=2, Pending=9, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 8\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 20\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout configured\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout configured\n- malware.antivirus | Antivirus | Non-Compliant | Windows Defender (off, out of date)\n- malware.service | Microsoft Defender Antimalware | Non-Compliant | service True, antivirus False\n- malware.realtime | Real-time protection | Non-Compliant | off\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Non-Compliant | off\n- malware.signatures | Security intelligence | Compliant | 2 days old (version 1.443.414.0)\n- firewall.domain.enabled | Domain firewall | Non-Compliant | off\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Non-Compliant | block\n- firewall.private.enabled | Private firewall | Non-Compliant | off\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Non-Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | allow\n- screen_lock.enabled | Screen lock | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.password | Screen lock password requirement | Error | not read\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an enabled administrator\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | LAPTOP-ZRCPWCK\\akaya\n- user.account_type | Signed-in user account type | Compliant | Standard (CONTOSO\\lmuller)\n- encryption.feature | BitLocker | Non-Compliant | off\n- encryption.enabled | Encryption | Non-Compliant | EncryptionInProgress (42%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, None\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 3 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 570 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 6 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Enterprise build 26100\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: success; account management: none\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | File History (jdoe)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | winrm, chromeremotedesktop\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=0 (policy)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Claude Code\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.expiration: Low\n- password.min_age: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.antivirus: High\n- malware.service: High\n- malware.realtime: High\n- malware.tamper: Medium\n- firewall.domain.enabled: High\n- firewall.domain.inbound: High\n- firewall.domain.outbound: Low\n- firewall.private.enabled: High\n- firewall.private.inbound: High\n- firewall.private.outbound: Low\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- encryption.feature: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.security_logging: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- org.breach_procedure: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=14, assessed=52, percent=27\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | Windows Defender (off, out of date)\n- malware.service | Microsoft Defender Antimalware | High | service True, antivirus False\n- malware.realtime | Real-time protection | High | off\n- firewall.domain.enabled | Domain firewall | High | off\n- firewall.domain.inbound | Domain inbound | High | allow\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\n- malware.service: Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\n- malware.realtime: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-ZRCPWCK - Microsoft Windows 11 Enterprise 10.0.26100\", \"score\": {\"compliant\": 14, \"assessed\": 52, \"percent\": 27}, \"counts\": {\"Compliant\": 14, \"Non-Compliant\": 38, \"NotApplicable\": 2, \"Pending\": 9, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"Windows Defender (off, out of date)\"}, {\"check_id\": \"malware.service\", \"name\": \"Microsoft Defender Antimalware\", \"risk\": \"High\", \"found\": \"service True, antivirus False\"}, {\"check_id\": \"malware.realtime\", \"name\": \"Real-time protection\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.enabled\", \"name\": \"Domain firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}], \"summary\": \"LAPTOP-ZRCPWCK meets 14 of 52 assessed checks (27%). The most serious gaps are Antivirus, Microsoft Defender Antimalware and Real-time protection. 1 checks could not be read. 9 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Antivirus: Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\", \"Microsoft Defender Antimalware: Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\", \"Real-time protection: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\", \"Answer the 6 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or use USBGuard.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read matches the required setting\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Malware signature updates\", \"finding\": \"The device reports freshclam not running, but the baseline requires automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted: does not meet the baseline\", \"finding\": \"Apple Intelligence external integrations restricted is not restricted instead of restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (windows):\n- Enable real-time protection in Microsoft Defender (on-access protection follows it).\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access protection: does not meet the baseline\", \"finding\": \"The device reports off, but the baseline requires on. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Enable real-time protection in Microsoft Defender (on-access protection follows it).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain outbound is non-compliant\", \"finding\": \"The device reports block, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"AI literacy training is partial instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Removable media encryption is USB storage writable without enforced encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Screen lock timeout is no machine inactivity limit; screen saver off instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: set automatically\nREQUIRED: set automatically\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Date & time set automatically: meets the baseline\", \"finding\": \"Date & time set automatically is set to set automatically, in line with the required set automatically. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone is non-compliant\", \"finding\": \"The user reports off, but the baseline requires on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Find My iPhone\", \"finding\": \"Find My iPhone is off instead of on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Unauthorized employee administrator accounts is jdoe instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength: does not meet the baseline\", \"finding\": \"The user reports 4-digit PIN, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Found: allow (no firewall). Required: deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption protection is non-compliant\", \"finding\": \"Disk encryption protection is not encrypted instead of active, strong cipher. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"Diagnostic data minimised is share Mac analytics=1, share with app developers=0 instead of analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Grok\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Grok. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Compliant\", \"title\": \"Private inbound is compliant\", \"finding\": \"Found: block. This satisfies the baseline requirement of block.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAI TOOLS FOUND: Jan, Hugging Face models\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /home/jdoe/.cache/huggingface/hub (37G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Jan: Local model runner; provider Open source (runs locally); data: On the device unless a cloud model is connected; default risk minimal; GDPR: Check whether remote API providers are configured; if so, the cloud provider rules apply.\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Jan\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Check whether remote API providers are configured; if so, the cloud provider rules apply.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Jan, Hugging Face models.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period: meets the baseline\", \"finding\": \"Lockout observation period is set to 30 minutes, in line with the required >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Private outbound: does not meet the baseline\", \"finding\": \"Found: allow. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google backup\", \"finding\": \"Google backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamonacc running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Compliant\", \"title\": \"On-access malware protection matches the required setting\", \"finding\": \"The device reports clamonacc running, which meets the requirement (on-access scanning).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: <= 0 pending\nDETAIL: Apple's update service could not be reached.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Error\", \"title\": \"Pending security updates: could not be checked\", \"finding\": \"The collector could not read this setting (Apple's update service could not be reached.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"The device reports never, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.7 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates is non-compliant\", \"finding\": \"iOS security updates is iOS 26.7 (update to 26.7.1 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nHOSTNAME: legal-lt-87\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=31, Non-Compliant=20, NotApplicable=2, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 99999\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | pam_pwquality not enabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Compliant | 2 days old\n- malware.on_access | On-access malware protection | Compliant | clamonacc running\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: sali (0s)\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | akaya, asmith\n- user.account_type | Signed-in user account type | Compliant | Standard (sali)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 40 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 175 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | LM Studio, Alpaca (Ollama client), Jan\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | LM Studio, Alpaca (Ollama client), Jan\n- aiact.risk_classification | AI system risk classification | Pending | LM Studio, Alpaca (Ollama client), Jan\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- password.complexity: Medium\n- malware.signature_updates: Medium\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- aiact.unapproved: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=31, assessed=51, percent=61\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee administrator accounts | High | akaya, asmith\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 40 critical/high\n- gdpr.updates_recent | Security updates installed recently | High | 175 days ago\n- gdpr.os_supported | Supported operating system version | High | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.backup_configured | Backup configured | High | none\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.updates_recent: Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors."}, {"role": "assistant", "content": "{\"device\": \"legal-lt-87 - Red Hat Enterprise Linux 8.10 (Ootpa) 8.10\", \"score\": {\"compliant\": 31, \"assessed\": 51, \"percent\": 61}, \"counts\": {\"Compliant\": 31, \"Non-Compliant\": 20, \"NotApplicable\": 2, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"akaya, asmith\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"40 critical/high\"}, {\"check_id\": \"gdpr.updates_recent\", \"name\": \"Security updates installed recently\", \"risk\": \"High\", \"found\": \"175 days ago\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Red Hat Enterprise Linux 8.10 (Ootpa)\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}], \"summary\": \"legal-lt-87 meets 31 of 51 assessed checks (61%). The most serious gaps are Unauthorized employee administrator accounts, Known vulnerabilities in installed software and Security updates installed recently. 3 items are pending.\", \"next_steps\": [\"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Security updates installed recently: Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Backup configured is none found instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 99999\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"The device reports PASS_MAX_DAYS 99999, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"Diagnostic data minimised is share Mac analytics=1, share with app developers=0 instead of analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"AI literacy training is partial instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Account lockout duration on Linux: Account lockout duration on Linux. Why it matters: The lockout must last long enough to slow down repeated guessing. Risk if missing: Low - A short lockout lets attackers resume guessing almost at once. Requirement: Locked for >= LOCKOUT_DURATION_MIN_MINUTES minutes, or until an administrator unlocks. How to fix on Linux: (1) Set unlock_time = 900 (seconds) in /etc/security/faillock.conf. How to verify: Re-run the audit; Account lockout duration should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Account lockout threshold on Linux: Account lockout threshold on Linux. Why it matters: Locking an account after a few wrong passwords stops online guessing. Risk if missing: Medium - Unlimited attempts allow brute-force attacks against local accounts. Requirement: Lockout after 1..LOCKOUT_THRESHOLD_MAX failed attempts (0 = never locks). How to fix on Linux: (1) Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)). (2) Set deny = 5 in /etc/security/faillock.conf. How to verify: Re-run the audit; Account lockout threshold should be between 1 and 5. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Lockout observation period on Linux: Lockout observation period on Linux. Why it matters: Failed attempts must be counted over a long enough window to detect slow guessing. Risk if missing: Low - A short window lets attackers pace their guesses to avoid lockout. Requirement: Failed-attempt counter kept for >= LOCKOUT_OBSERVATION_MIN_MINUTES minutes. How to fix on Linux: (1) Set fail_interval = 900 (seconds) in /etc/security/faillock.conf. How to verify: Re-run the audit; Lockout observation period should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Account lockout duration on Windows: Account lockout duration on Windows. Why it matters: The lockout must last long enough to slow down repeated guessing. Risk if missing: Low - A short lockout lets attackers resume guessing almost at once. Requirement: Locked for >= LOCKOUT_DURATION_MIN_MINUTES minutes, or until an administrator unlocks. How to fix on Windows: (1) Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more. (2) For a stand-alone device: net accounts /lockoutduration:15 How to verify: Re-run the audit; Account lockout duration should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Ubuntu 26.04.1 LTS 26.04 (linux), audited 2026-10-02; 13 of 47 assessed checks compliant; 34 gaps, 5 pending, 0 not read. Gaps: Password history: 5; Password expiration: PASS_MAX_DAYS 99999; Password complexity: pam_pwquality not enabled; Account lockout threshold: 10; Account lockout duration: 10 minutes; Lockout observation period: no lockout; ClamAV malware protection service: not running; Malware signature updates: freshclam not running.\n\nQUESTION: How do I fix account lockout duration?"}, {"role": "assistant", "content": "Your audit found Account lockout duration: 10 minutes. The lockout must last long enough to slow down repeated guessing. To fix it [1]:\n1. Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "linux", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nHOSTNAME: DESKTOP-UVFE9AD\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=46, Non-Compliant=9, NotApplicable=8, Pending=0, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | Bitdefender Endpoint Security (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | password-protected screen saver on for all signed-in users\n- screen_lock.timeout | Screen lock timeout | Compliant | 180 seconds (machine inactivity limit)\n- screen_lock.password | Screen lock password requirement | Compliant | machine inactivity limit locks the session\n- usb.device_access | USB storage device access | Non-Compliant | deny\n- usb.read | USB storage read | Non-Compliant | deny\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | CONTOSO\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (CONTOSO\\pmartin)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 1 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 21 days ago\n- gdpr.os_supported | Supported operating system version | Error | not read\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=0 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Compliant | Recall disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- usb.device_access: Low\n- usb.read: Low\n- user.account_type: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- org.processor_agreements: High\n- org.ai_register: Medium\nSCORE (computed): compliant=46, assessed=55, percent=84\nTOP GAPS (computed, highest risk first):\n- user.account_type | Signed-in user account type | High | Administrator (CONTOSO\\pmartin)\n- vuln.pending_updates | Pending security updates | High | 2 pending\n- gdpr.backup_configured | Backup configured | High | none found\n- org.processor_agreements | Data processing agreements | High | partial\n- vuln.outdated_apps | Outdated third-party applications | Medium | 1 outdated\nFIX GUIDANCE (highest-risk gaps):\n- user.account_type: Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\n- vuln.pending_updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\n- gdpr.backup_configured: Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-UVFE9AD - Microsoft Windows 11 Enterprise 10.0.26100\", \"score\": {\"compliant\": 46, \"assessed\": 55, \"percent\": 84}, \"counts\": {\"Compliant\": 46, \"Non-Compliant\": 9, \"NotApplicable\": 8, \"Pending\": 0, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (CONTOSO\\\\pmartin)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"2 pending\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none found\"}, {\"check_id\": \"org.processor_agreements\", \"name\": \"Data processing agreements\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"vuln.outdated_apps\", \"name\": \"Outdated third-party applications\", \"risk\": \"Medium\", \"found\": \"1 outdated\"}], \"summary\": \"DESKTOP-UVFE9AD meets 46 of 55 assessed checks (84%). The most serious gaps are Signed-in user account type, Pending security updates and Backup configured. 1 checks could not be read.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Signed-in user account type: Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\", \"Pending security updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\", \"Backup configured: Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAI TOOLS FOUND: GPT4All\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- GPT4All: Local model runner; provider Nomic AI (runs locally); data: On the device unless optional data sharing is enabled; default risk minimal; GDPR: Keep the optional data lake sharing switched off.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"GPT4All\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Keep the optional data lake sharing switched off.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: GPT4All.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history: meets the baseline\", \"finding\": \"The device reports 24, which meets the requirement (>= 24).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: GPT4All, Ollama, Alpaca (Ollama client)\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Found: GPT4All, Ollama, Alpaca (Ollama client). Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: rschmit (1800s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: rschmit\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Screen lock timeout is too long for: rschmit (1800s) instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker recovery\", \"finding\": \"Found: key protectors: Tpm. Required: recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 14 (android)\nAI TOOLS FOUND: ChatGPT, Meta AI\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Meta AI: Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps); provider Meta (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Meta AI\", \"kind\": \"Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Meta AI.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Data protection impact assessment is partial instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 4\nREQUIRED: 1-5\nDETAIL: pam_faillock\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Compliant\", \"title\": \"Account lockout threshold matches the required setting\", \"finding\": \"The device reports 4, which meets the requirement (1-5).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: minclass=0, required classes=0\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"The device reports minclass=0, required classes=0, but the baseline requires 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 15.7.2 (macos)\nAI TOOLS FOUND: Jan, Hugging Face models, Cursor\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /Users/jdoe/.cache/huggingface/hub (19G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Jan: Local model runner; provider Open source (runs locally); data: On the device unless a cloud model is connected; default risk minimal; GDPR: Check whether remote API providers are configured; if so, the cloud provider rules apply.\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\n- Cursor: AI code editor; provider Anysphere (US); data: Code sent to the provider and its model providers unless privacy mode is enforced; default risk minimal; GDPR: Enforce privacy mode through a business plan and sign a data processing agreement.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Jan\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Check whether remote API providers are configured; if so, the cloud provider rules apply.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Cursor\", \"kind\": \"AI code editor\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Enforce privacy mode through a business plan and sign a data processing agreement.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Jan, Hugging Face models, Cursor.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: NTP service yes, synchronised yes\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation: meets the baseline\", \"finding\": \"Time synchronisation is set to NTP service yes, synchronised yes, in line with the required time synchronised.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"ClamAV malware protection service: does not meet the baseline\", \"finding\": \"ClamAV malware protection service is not running instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] ISO 27001:2022 A.8.20 Networks security: ISO 27001:2022 A.8.20 (Networks security): Secure, manage and control networks and network devices to protect information. An auditor looks for: Host firewalls enabled; Unsolicited inbound connections blocked. On endpoints: The host firewall protects the laptop on untrusted networks such as hotels and cafés.\n[2] BitLocker on Windows: BitLocker on Windows. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: BitLocker is turned on for the system drive. How to fix on Windows: (1) Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[3] Encryption on Windows: Encryption on Windows. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: The system disk is fully encrypted (BitLocker / LUKS / FileVault). Phones: iOS data protection is on when a passcode is set; Android 10+ encrypts by default. How to fix on Windows: (1) Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[4] ISO 27001:2022 A.5.12 Classification of information: ISO 27001:2022 A.5.12 (Classification of information): Label information by how sensitive it is so that the right protection can be applied. An auditor looks for: A classification scheme; Evidence that sensitive data is identified, e.g. personal data. On endpoints: Personal or confidential files stored on laptops need to be found and protected accordingly.\n\nMY AUDIT: none\n\nQUESTION: What's the best gaming laptop?"}, {"role": "assistant", "content": "I can only help with device security, ISO 27001, GDPR and the EU AI Act, so I can't answer that. Ask me, for example, how to turn on disk encryption, whether an AI tool is OK for work, or run an audit of this device."}], "meta": {"task": "chat", "os": "any", "status": "oos", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"AI transparency to users is partial instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 12\nREQUIRED: >= Android 13\nDETAIL: Version entered by the user from Settings › About phone. This version no longer receives all security updates.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"The user reports Android 12, but the baseline requires >= Android 13. An operating system that no longer receives security updates cannot be kept secure. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Homebrew packages checked; App Store apps are not covered.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (macos):\n- Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Compliant\", \"title\": \"Outdated third-party applications: meets the baseline\", \"finding\": \"Found: 0 outdated. This satisfies the baseline requirement of <= 0 outdated.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Private inbound\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (windows):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not applicable\", \"finding\": \"Not applicable on this device: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamav-freshclam\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Compliant\", \"title\": \"Malware signature updates is compliant\", \"finding\": \"Found: clamav-freshclam. This satisfies the baseline requirement of automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities: does not meet the baseline\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Removable media encryption is external drives writable without enforced encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: ssh, screensharing\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Found: ssh, screensharing. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Sensitive notifications hidden when locked\", \"finding\": \"Sensitive notifications hidden when locked is all content shown instead of sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates: does not meet the baseline\", \"finding\": \"Found: manual. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 11 Pro build 22621\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"Found: Microsoft Windows 11 Pro build 22621. Required: build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup: does not meet the baseline\", \"finding\": \"The user reports no backup, but the baseline requires backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: within limit\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: lmuller, asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout is compliant\", \"finding\": \"Found: within limit. This satisfies the baseline requirement of <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Found: off. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public outbound\", \"finding\": \"Found: block. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 39 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 22 applications inventoried, 11 recognised (23 common applications covered). Visual Studio Code 1.123.1 (21 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-40376, CVE-2026-45482); zoom.us 6.5.5 (6 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-62483, CVE-2025-64739); iTerm 3.4.18 (5 CVEs, critical; fixed in 3.4.20; e.g. CVE-2023-46300, CVE-2023-46301); Safari 26.5.2 (7 CVEs, high; fixed in 26.6.1; e.g. CVE-2026-43794, CVE-2026-64713). 35 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"The device reports 39 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Copilot\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The user reports Microsoft Copilot, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker recovery\", \"finding\": \"BitLocker recovery is key protectors: Tpm instead of recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: pmartin, mkhan\nREQUIRED: password required\nDETAIL: Users checked: pmartin, mkhan\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"The device reports not required for: pmartin, mkhan, but the baseline requires password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Character.AI\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Character.AI. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: chromeremotedesktop\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"Remote access services restricted is chromeremotedesktop instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.7 (ios)\nHOSTNAME: HR-iPhone-10\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=19, Non-Compliant=6, NotApplicable=0, Pending=4, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 5 minutes\n- mobile.passcode_strength | Passcode strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store only\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.7 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 15 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | Character.AI\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Character.AI\n- aiact.risk_classification | AI system risk classification | Pending | Character.AI\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Pending | not answered\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- aiact.unapproved: Medium\n- org.ropa: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=19, assessed=25, percent=76\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | iOS security updates | High | iOS 26.7 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 15 critical/high\n- org.ropa | Record of processing activities | High | partial\n- vuln.outdated_apps | App updates | Medium | automatic app updates off\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | Character.AI\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- org.ropa: Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures)."}, {"role": "assistant", "content": "{\"device\": \"HR-iPhone-10 - iOS 26.7\", \"score\": {\"compliant\": 19, \"assessed\": 25, \"percent\": 76}, \"counts\": {\"Compliant\": 19, \"Non-Compliant\": 6, \"NotApplicable\": 0, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.7 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"15 critical/high\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"vuln.outdated_apps\", \"name\": \"App updates\", \"risk\": \"Medium\", \"found\": \"automatic app updates off\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"Character.AI\"}], \"summary\": \"HR-iPhone-10 meets 19 of 25 assessed checks (76%). The most serious gaps are iOS security updates, Known vulnerabilities in iOS and Record of processing activities. 4 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Record of processing activities: Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"Found: 0 days. Required: >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Data processing agreements is no instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"Firewall is no active firewall instead of active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"Found: no lockout configured. Required: >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Passcode strength\", \"finding\": \"The user reports 4-digit code, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 18.7 (update to 18.7.10 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 18.7.10 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates: does not meet the baseline\", \"finding\": \"Found: iOS 18.7 (update to 18.7.10 available). Required: latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 24 applications inventoried, 13 recognised (31 common applications covered). 0 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Compliant\", \"title\": \"Known vulnerabilities in installed software: meets the baseline\", \"finding\": \"Known vulnerabilities in installed software is set to 0 critical/high, in line with the required <= 0 critical/high.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no passcode\nREQUIRED: passcode set\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Passcode: does not meet the baseline\", \"finding\": \"Passcode is no passcode instead of passcode set. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iOS 26.7\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Safari 26.7).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version: meets the baseline\", \"finding\": \"The device reports iOS 26.7, which meets the requirement (>= iOS 18).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device is non-compliant\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Security intelligence on Windows: Security intelligence on Windows. Why it matters: Anti-malware can only detect threats it knows about; signatures must be current. Risk if missing: High - Outdated signatures miss recent malware. Requirement: Signatures no older than SIGNATURE_MAX_AGE_DAYS (macOS XProtect: MACOS_XPROTECT_MAX_AGE_DAYS). How to fix on Windows: (1) Run Update-MpSignature and check that the device can reach Windows Update or your update server. How to verify: Re-run the audit; signatures should be within the allowed age. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Screen lock on Windows: Screen lock on Windows. Why it matters: An idle device should lock itself so nobody can use it while the owner is away. Risk if missing: Medium - An unlocked, unattended device exposes all the user's data and accounts. Requirement: The screen locks when idle (Windows: machine inactivity limit or a password-protected screen saver). Phones: a passcode / PIN / password screen lock is set. How to fix on Windows: (1) Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune. How to verify: Re-run the audit; Screen lock should be on. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Antivirus on Windows: Antivirus on Windows. Why it matters: Anti-malware software detects and blocks known malicious files and behaviour. Risk if missing: High - Without working anti-malware, ransomware and info-stealers can run unchallenged. Requirement: An enabled, up-to-date antivirus product is present. Android: Google Play Protect scanning is on. How to fix on Windows: (1) Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center. How to verify: Re-run the audit; Antivirus should be Compliant. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Windows Recall / Copilot data analysis disabled on Windows: Windows Recall / Copilot data analysis disabled on Windows. Why it matters: Built-in OS AI features can capture screen content or send text to external AI services. Risk if missing: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose. Requirement: OS-level AI features that capture or send user data are disabled by policy. NotApplicable where the OS has none. Phones: external AI integrations (ChatGPT in Apple Intelligence, assistant access to screen content) are off unless approved. How to fix on Windows: (1) Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved. How to verify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n\nMY AUDIT: none\n\nMY DEVICE: Windows 11 (windows) on a Windows PC (x86-64), Edge 141; detected by the browser, security settings not checked.\n\nQUESTION: How can I set up security intelligence on my Windows?"}, {"role": "assistant", "content": "Anti-malware can only detect threats it knows about; signatures must be current. To fix it on Windows [1]:\n1. Run Update-MpSignature and check that the device can reach Windows Update or your update server.\n\nTo check it worked: Re-run the audit; signatures should be within the allowed age."}], "meta": {"task": "chat", "os": "windows", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"The device reports 12, but the baseline requires >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: automatic\nREQUIRED: automatic\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic system updates is compliant\", \"finding\": \"Automatic system updates is set to automatic, in line with the required automatic. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"Lockout observation period is 10 minutes instead of >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nDETAIL: removable disk write denied by policy\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Compliant\", \"title\": \"USB storage write is compliant\", \"finding\": \"USB storage write is set to deny, in line with the required deny.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 18.7 (ios)\nHOSTNAME: HR-iPhone-47\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=8, Non-Compliant=15, NotApplicable=0, Pending=6, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 18.7 (update to 18.7.10 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 207 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | download only\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 18.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Claude\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Claude\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Claude\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=8, assessed=23, percent=35\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- vuln.pending_updates | iOS security updates | High | iOS 18.7 (update to 18.7.10 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 207 critical/high\n- gdpr.auto_updates | Automatic iOS updates | High | download only\n- gdpr.backup_configured | iCloud Backup | High | no backup\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"HR-iPhone-47 - iOS 18.7\", \"score\": {\"compliant\": 8, \"assessed\": 23, \"percent\": 35}, \"counts\": {\"Compliant\": 8, \"Non-Compliant\": 15, \"NotApplicable\": 0, \"Pending\": 6, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 18.7 (update to 18.7.10 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"207 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"download only\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}], \"summary\": \"HR-iPhone-47 meets 8 of 23 assessed checks (35%). The most serious gaps are Apps only from the App Store, iOS security updates and Known vulnerabilities in iOS. 6 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 4 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Found: 6. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Data protection impact assessment is partial instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"The device reports never, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: allow\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Error\", \"title\": \"Private outbound: could not be checked\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (windows):\n- Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read matches the required setting\", \"finding\": \"USB storage read is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 26.6 (ios)\nAI TOOLS FOUND: ChatGPT, Grammarly\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Grammarly: AI writing assistant (keyboard and browser); provider Grammarly (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Reads everything typed where it is enabled, including personal data; use the business plan with a data processing agreement or disable it in work apps.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Grammarly\", \"kind\": \"AI writing assistant (keyboard and browser)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Reads everything typed where it is enabled, including personal data; use the business plan with a data processing agreement or disable it in work apps.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Grammarly.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"Data processing agreements is partial instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: guest login off\nREQUIRED: guest login off\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (macos):\n- sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled: meets the baseline\", \"finding\": \"Guest account disabled is set to guest login off, in line with the required guest login off.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: ClamAV malware protection service\", \"finding\": \"The device reports not running, but the baseline requires running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-1V6Y5K3\\ewagner\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts: does not meet the baseline\", \"finding\": \"Unauthorized employee local administrator accounts is DESKTOP-1V6Y5K3\\\\ewagner instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"Found: not set. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private firewall [firewall.private.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Private firewall\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in bzserv.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: decision pending\", \"finding\": \"Confirm the last backup date in bzserv. Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 180 idle, password delay 300\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Screen lock timeout is 180 idle, password delay 300 instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery is non-compliant\", \"finding\": \"Found: key protectors: Tpm. Required: recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Public inbound is non-compliant\", \"finding\": \"Public inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Found: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: active\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Error\", \"title\": \"Firewall: could not be checked\", \"finding\": \"This check returned an error: Requires root: run the collector with sudo. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nHOSTNAME: DESKTOP-6HPD90X\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=49, Non-Compliant=5, NotApplicable=8, Pending=2, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | Sophos Endpoint (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Compliant | 300 seconds (machine inactivity limit)\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | DEVSEIS\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (DEVSEIS\\rschmit)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 13 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 24 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 11 Pro build 22621\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | OneDrive folder backup (policy)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=1 (policy)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Compliant | Recall disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- password.complexity: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- org.ropa: High\nSCORE (computed): compliant=49, assessed=54, percent=91\nTOP GAPS (computed, highest risk first):\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 13 critical/high\n- gdpr.os_supported | Supported operating system version | High | Microsoft Windows 11 Pro build 22621\n- org.ropa | Record of processing activities | High | partial\n- password.complexity | Password complexity | Medium | disabled\n- password.expiration | Password expiration | Low | never\nFIX GUIDANCE (highest-risk gaps):\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.os_supported: Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\n- org.ropa: Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures)."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-6HPD90X - Microsoft Windows 11 Pro 10.0.22621\", \"score\": {\"compliant\": 49, \"assessed\": 54, \"percent\": 91}, \"counts\": {\"Compliant\": 49, \"Non-Compliant\": 5, \"NotApplicable\": 8, \"Pending\": 2, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"13 critical/high\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Microsoft Windows 11 Pro build 22621\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"password.complexity\", \"name\": \"Password complexity\", \"risk\": \"Medium\", \"found\": \"disabled\"}, {\"check_id\": \"password.expiration\", \"name\": \"Password expiration\", \"risk\": \"Low\", \"found\": \"never\"}], \"summary\": \"DESKTOP-6HPD90X meets 49 of 54 assessed checks (91%). The most serious gaps are Known vulnerabilities in installed software, Supported operating system version and Record of processing activities. 2 items are pending.\", \"next_steps\": [\"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Supported operating system version: Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\", \"Record of processing activities: Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI register\", \"finding\": \"The device reports no, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection is non-compliant\", \"finding\": \"On-access malware protection is not running instead of on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates is non-compliant\", \"finding\": \"Found: automatic app updates off. Required: automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (windows):\n- Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read is compliant\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 26.0.1 (macos)\nAI TOOLS FOUND: Cursor\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Cursor: AI code editor; provider Anysphere (US); data: Code sent to the provider and its model providers unless privacy mode is enforced; default risk minimal; GDPR: Enforce privacy mode through a business plan and sign a data processing agreement.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Cursor\", \"kind\": \"AI code editor\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Enforce privacy mode through a business plan and sign a data processing agreement.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Cursor.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect 5360\nREQUIRED: antivirus present\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (macos):\n- XProtect is built in; if it is missing or damaged, reinstall the latest macOS update. Add a managed endpoint protection product if your policy requires on-access scanning.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus is compliant\", \"finding\": \"Antivirus is set to XProtect 5360, in line with the required antivirus present.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Claude Code (AI tool): Claude Code: AI coding assistant (command line), provider Anthropic (US), data location: Code and prompts sent to the provider's cloud. Default EU AI Act risk class for general office use: minimal. GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[2] Claude (AI tool): Claude: Cloud AI assistant, provider Anthropic (US), data location: Provider cloud outside the EEA. Default EU AI Act risk class for general office use: minimal. GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44). Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[3] Windsurf (AI tool): Windsurf: AI code editor, provider US-based provider, data location: Code sent to the provider's cloud. Default EU AI Act risk class for general office use: minimal. GDPR: Use a business plan with a data processing agreement and zero-retention settings. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[4] Cursor (AI tool): Cursor: AI code editor, provider Anysphere (US), data location: Code sent to the provider and its model providers unless privacy mode is enforced. Default EU AI Act risk class for general office use: minimal. GDPR: Enforce privacy mode through a business plan and sign a data processing agreement. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n\nMY AUDIT: none\n\nQUESTION: What is the EU AI Act risk class of Claude Code?"}, {"role": "assistant", "content": "Claude Code is a ai coding assistant (command line) from Anthropic (US); data goes to: code and prompts sent to the provider's cloud [1]. For general office use its default EU AI Act risk class is minimal. GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\n\nWhether it is allowed depends on your organisation's approved AI tools list and on what you use it for: some purposes (for example screening job applicants) make any AI tool high-risk."}], "meta": {"task": "chat", "os": "any", "status": "tool", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 10 Pro build 19045\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"Found: Microsoft Windows 10 Pro build 19045. Required: build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private firewall [firewall.private.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Private firewall\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI register\", \"finding\": \"Found: partial. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Standard (FABRIKAM\\sali)\nREQUIRED: Standard\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Compliant\", \"title\": \"Signed-in user account type: meets the baseline\", \"finding\": \"Found: Standard (FABRIKAM\\\\sali). This satisfies the baseline requirement of Standard.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 14 (android)\nHOSTNAME: sales-galaxy-34\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=6, Non-Compliant=14, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 0 detected by the browser, 22 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Non-Compliant | no screen lock (Swipe/None)\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 15 seconds\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | swipe or none\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Non-Compliant | on\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 14 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2025-03 (19 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Compliant | Android 14\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | ChatGPT\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.enabled: Medium\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.usb_debugging: Medium\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=6, assessed=20, percent=30\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- vuln.pending_updates | Android security update | High | security patch 2025-03 (19 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.backup_configured | Google backup | High | no backup\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone."}, {"role": "assistant", "content": "{\"device\": \"sales-galaxy-34 - Android 14\", \"score\": {\"compliant\": 6, \"assessed\": 20, \"percent\": 30}, \"counts\": {\"Compliant\": 6, \"Non-Compliant\": 14, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2025-03 (19 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Google backup\", \"risk\": \"High\", \"found\": \"no backup\"}], \"summary\": \"sales-galaxy-34 meets 6 of 20 assessed checks (30%). The most serious gaps are Google Play Protect, Install unknown apps blocked and Android security update. 10 items are pending. 22 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock is non-compliant\", \"finding\": \"The user reports Auto-Lock Never, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"Found: 5 minutes. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutwindow:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no screen lock (Swipe/None)\nREQUIRED: PIN, password or pattern\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"Screen lock is no screen lock (Swipe/None) instead of PIN, password or pattern. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Cursor, Ollama, Claude\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: Cursor, Ollama, Claude. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: download and install on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Pending\", \"title\": \"Automatic iOS updates is pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rooted or bootloader unlocked\nREQUIRED: not rooted\nDETAIL: Android security protections are weakened on a rooted phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not rooted is non-compliant\", \"finding\": \"Not rooted is rooted or bootloader unlocked instead of not rooted. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates is non-compliant\", \"finding\": \"Found: check=1 download=1 security=1 macOS-install=0. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"AI literacy training is partial instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAI TOOLS FOUND: LM Studio\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- LM Studio: Local model runner; provider Element Labs (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences and whether its local server is exposed.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"LM Studio\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences and whether its local server is exposed.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: LM Studio.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Compliant\", \"title\": \"Unauthorized employee administrator accounts is compliant\", \"finding\": \"Found: none. This satisfies the baseline requirement of none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 13 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 19 applications inventoried, 11 recognised (31 common applications covered). Notepad++ (64-bit x64) 8.4.5 (5 CVEs, high, actively exploited; fixed in 8.9.6.1; e.g. CVE-2023-40031, CVE-2025-15556); AnyDesk 5.5.3 (8 CVEs, critical; fixed in 9.0.0; e.g. CVE-2020-27614, CVE-2020-35483). 6 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Found: 13 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Gemini\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: Gemini. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: download and install on\nDETAIL: Security updates are not installed automatically.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic iOS updates: does not meet the baseline\", \"finding\": \"Automatic iOS updates is off instead of download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Hugging Face models, Alpaca (Ollama client)\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: Hugging Face models, Alpaca (Ollama client). Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private firewall [firewall.private.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Private firewall is non-compliant\", \"finding\": \"Private firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"The device reports none, which meets the requirement (only: none).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: File History (jdoe).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Found: date not readable. Confirm the last backup date in: File History (jdoe).\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"The device reports 0, but the baseline requires >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\", \"For a stand-alone device: net accounts /uniquepw:24\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: chromeremotedesktop\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"Remote access services restricted is chromeremotedesktop instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: ClamAV (clamav-daemon)\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus: meets the baseline\", \"finding\": \"The device reports ClamAV (clamav-daemon), which meets the requirement (antivirus running).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic system updates\", \"finding\": \"Automatic system updates is manual instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"Found: none running. Required: antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Breach response procedure is partial instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.4 (ios)\nHOSTNAME: SALES-iPhone-36\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=8, Non-Compliant=9, NotApplicable=3, Pending=9, Error=0\nEVIDENCE: 3 detected by the browser, 18 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Non-Compliant | no passcode\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 4 minutes\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Non-Compliant | jailbroken\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Non-Compliant | off (no passcode)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.4 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Pending | not answered\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 142 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.4\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | When Unlocked\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- encryption.enabled: High\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.time_sync: Low\nSCORE (computed): compliant=8, assessed=17, percent=47\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not jailbroken | High | jailbroken\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- encryption.enabled | Data protection (encryption) | High | off (no passcode)\n- vuln.pending_updates | iOS security updates | High | iOS 26.4 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 142 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- encryption.enabled: Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set."}, {"role": "assistant", "content": "{\"device\": \"SALES-iPhone-36 - iOS 26.4\", \"score\": {\"compliant\": 8, \"assessed\": 17, \"percent\": 47}, \"counts\": {\"Compliant\": 8, \"Non-Compliant\": 9, \"NotApplicable\": 3, \"Pending\": 9, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not jailbroken\", \"risk\": \"High\", \"found\": \"jailbroken\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Data protection (encryption)\", \"risk\": \"High\", \"found\": \"off (no passcode)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.4 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"142 critical/high\"}], \"summary\": \"SALES-iPhone-36 meets 8 of 17 assessed checks (47%). The most serious gaps are Not jailbroken, Apps only from the App Store and Data protection (encryption). 9 items are pending. 18 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not jailbroken: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\", \"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Data protection (encryption): Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (windows):\n- Enable real-time protection in Microsoft Defender (on-access protection follows it).\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access protection is non-compliant\", \"finding\": \"On-access protection is off instead of on. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Enable real-time protection in Microsoft Defender (on-access protection follows it).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 26.7.1 (ios)\nAI TOOLS FOUND: ChatGPT, Perplexity\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Perplexity: Cloud AI search assistant; provider Perplexity (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs an enterprise agreement and a transfer basis before work or personal data is entered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Perplexity\", \"kind\": \"Cloud AI search assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs an enterprise agreement and a transfer basis before work or personal data is entered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Perplexity.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (DESKTOP-J0Q0Y2I\\ewagner)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"The device reports Administrator (DESKTOP-J0Q0Y2I\\\\ewagner), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamav-freshclam\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Compliant\", \"title\": \"Malware signature updates is compliant\", \"finding\": \"Found: clamav-freshclam. This satisfies the baseline requirement of automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"Account lockout duration is 10 minutes instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted is non-compliant\", \"finding\": \"Apple Intelligence external integrations restricted is not restricted instead of restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Compliant\", \"title\": \"Private inbound: meets the baseline\", \"finding\": \"Found: block. This satisfies the baseline requirement of block.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Pending\", \"title\": \"Find My Device: decision pending\", \"finding\": \"The user could not find or confirm this setting. Current state: not answered. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1989 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 300 packages checked against rhel:9 advisories. kernel-64k-debug-devel-matched 0:5.14.0-427.24.1.el9_4 (1965 CVEs, high, actively exploited; fixed in 0:5.14.0-687.54.1.el9_8; e.g. CVE-2019-25162, CVE-2020-10135); mariadb-server-utils-debuginfo 3:10.5.22-1.el9_2 (15 CVEs, critical; fixed in 3:10.11.18-1.module+el9.8.0+24388+85f1c439; e.g. CVE-2023-52969, CVE-2023-52970); curl-minimal-debuginfo 0:7.76.1-23.el9 (9 CVEs, high; fixed in 0:7.76.1-40.el9_8.7; e.g. CVE-2023-38545, CVE-2023-38546). 32 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 1989 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nHOSTNAME: LAPTOP-6KE1NN4\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=36, Non-Compliant=20, NotApplicable=4, Pending=3, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Non-Compliant | 5\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 20\n- lockout.duration | Account lockout duration | Non-Compliant | 5 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Compliant | on\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Compliant | on\n- malware.signatures | Security intelligence | Compliant | 3 days old (version 1.444.453.0)\n- firewall.domain.enabled | Domain firewall | Non-Compliant | off\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Non-Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | password-protected screen saver on for all signed-in users\n- screen_lock.timeout | Screen lock timeout | Compliant | screen saver within limit for: FABRIKAM\\sali\n- screen_lock.password | Screen lock password requirement | Compliant | machine inactivity limit locks the session\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | FABRIKAM\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | LAPTOP-6KE1NN4\\asmith, LAPTOP-6KE1NN4\\Administrator\n- user.account_type | Signed-in user account type | Compliant | Standard (FABRIKAM\\sali)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 13 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | automatic updates turned off by policy\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 57 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26200\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: none; account management: none\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | OneDrive folder backup (policy)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Error | not read\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | chromeremotedesktop, anydesk\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=1 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.min_age: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- firewall.domain.enabled: High\n- firewall.domain.inbound: High\n- firewall.domain.outbound: Low\n- firewall.private.inbound: High\n- usb.write: Medium\n- admin.unauthorized: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- org.dpia: Medium\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=36, assessed=56, percent=64\nTOP GAPS (computed, highest risk first):\n- firewall.domain.enabled | Domain firewall | High | off\n- firewall.domain.inbound | Domain inbound | High | allow\n- firewall.private.inbound | Private inbound | High | allow\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | LAPTOP-6KE1NN4\\asmith, LAPTOP-6KE1NN4\\Administrator\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 13 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- firewall.domain.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- firewall.private.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-6KE1NN4 - Microsoft Windows 11 Pro 10.0.26200\", \"score\": {\"compliant\": 36, \"assessed\": 56, \"percent\": 64}, \"counts\": {\"Compliant\": 36, \"Non-Compliant\": 20, \"NotApplicable\": 4, \"Pending\": 3, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"firewall.domain.enabled\", \"name\": \"Domain firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"LAPTOP-6KE1NN4\\\\asmith, LAPTOP-6KE1NN4\\\\Administrator\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"13 critical/high\"}], \"summary\": \"LAPTOP-6KE1NN4 meets 36 of 56 assessed checks (64%). The most serious gaps are Domain firewall, Domain inbound and Private inbound. 1 checks could not be read. 3 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Domain firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Private inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"Account lockout duration is 10 minutes instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 15.7.2 (macos)\nAI TOOLS FOUND: Perplexity, ChatGPT, Claude\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Perplexity: Cloud AI search assistant; provider Perplexity (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs an enterprise agreement and a transfer basis before work or personal data is entered.\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Perplexity\", \"kind\": \"Cloud AI search assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs an enterprise agreement and a transfer basis before work or personal data is entered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Perplexity, ChatGPT, Claude.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"Lockout observation period is no lockout instead of >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 12\nREQUIRED: >= Android 13\nDETAIL: Version entered by the user from Settings › About phone. This version no longer receives all security updates.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"Found: Android 12. Required: >= Android 13. An operating system that no longer receives security updates cannot be kept secure. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 12\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Samsung Internet 28.0). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"Found: Android 12. Required: >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: tnguyen, ewagner\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"Unauthorized employee administrator accounts is tnguyen, ewagner instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"The device reports 5 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"Found: not encrypted. Required: recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Perplexity\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The user reports ChatGPT, Perplexity, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MIN_DAYS 0\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (linux):\n- Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"Found: PASS_MIN_DAYS 0. Required: >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: logon: success+failure; account management: success+failure\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging is compliant\", \"finding\": \"Found: logon: success+failure; account management: success+failure. This satisfies the baseline requirement of logon success+failure and account changes audited.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Google Play Protect: does not meet the baseline\", \"finding\": \"Found: Play Protect scanning off. Required: on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: LM Studio\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: LM Studio.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in bzserv.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Found: date not readable. Confirm the last backup date in bzserv.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"Human oversight for high-risk AI is partial instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: screen saver requires sign-in for all signed-in users\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement is compliant\", \"finding\": \"The device reports screen saver requires sign-in for all signed-in users, which meets the requirement (sign-in required to unlock).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nHOSTNAME: hr-lt-62\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=12, Non-Compliant=34, NotApplicable=7, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 10\n- password.history | Password history | Non-Compliant | 5\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Non-Compliant | minclass=0, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | not set\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Non-Compliant | 44 days old\n- malware.on_access | On-access malware protection | Compliant | clamonacc running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | NotApplicable | no GNOME desktop\n- screen_lock.timeout | Screen lock timeout | NotApplicable | no GNOME desktop\n- screen_lock.password | Screen lock password requirement | NotApplicable | no GNOME desktop\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | asmith\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (ewagner)\n- encryption.enabled | Encryption | Non-Compliant | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | not encrypted\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 1 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 142 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 26.04.1 LTS\n- gdpr.security_logging | Security event logging | Non-Compliant | auditd not running, journal volatile\n- gdpr.time_sync | Time synchronisation | Non-Compliant | NTP service no, synchronised no\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Non-Compliant | LightDM guest session allowed\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | smb\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.complexity: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.antivirus: High\n- malware.service: High\n- malware.signature_updates: Medium\n- malware.signatures: High\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=12, assessed=46, percent=26\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.service | ClamAV malware protection service | High | not running\n- malware.signatures | Malware signatures | High | 44 days old\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now."}, {"role": "assistant", "content": "{\"device\": \"hr-lt-62 - Ubuntu 26.04.1 LTS 26.04\", \"score\": {\"compliant\": 12, \"assessed\": 46, \"percent\": 26}, \"counts\": {\"Compliant\": 12, \"Non-Compliant\": 34, \"NotApplicable\": 7, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"44 days old\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}], \"summary\": \"hr-lt-62 meets 12 of 46 assessed checks (26%). The most serious gaps are Antivirus, ClamAV malware protection service and Malware signatures. 3 items are pending.\", \"next_steps\": [\"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Answer the 3 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"Found: no. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Found: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates is non-compliant\", \"finding\": \"Found: freshclam not running. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: App Store only\nREQUIRED: App Store only\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Compliant\", \"title\": \"Apps only from the App Store: meets the baseline\", \"finding\": \"Found: App Store only. This satisfies the baseline requirement of App Store only. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Compliant\", \"title\": \"Private outbound is compliant\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apple Intelligence external integrations restricted\", \"finding\": \"The device reports not restricted, but the baseline requires restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 900 seconds (machine inactivity limit)\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"The device reports 900 seconds (machine inactivity limit), but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Compliant\", \"title\": \"Domain outbound: meets the baseline\", \"finding\": \"Domain outbound is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: security responses and system data files install automatically\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Compliant\", \"title\": \"Malware security updates is compliant\", \"finding\": \"Malware security updates is set to security responses and system data files install automatically, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: Recall not available on this device\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Windows Recall / Copilot data analysis disabled: not applicable\", \"finding\": \"Windows Recall / Copilot data analysis disabled is not assessed here: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iPadOS 26.7 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iPadOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iOS security updates\", \"finding\": \"The user reports iPadOS 26.7 (update to 26.7.1 available), but the baseline requires latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Public inbound: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"AI literacy training is partial instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 106 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iPadOS 26.5 checked against NVD iOS records. iPadOS 26.5 (106 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-28928). 180 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in iOS\", \"finding\": \"Found: 106 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: download and install on\nREQUIRED: download and install on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic iOS updates matches the required setting\", \"finding\": \"Automatic iOS updates is set to download and install on, in line with the required download and install on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nHOSTNAME: DESKTOP-6F4VUQI\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=15, Non-Compliant=37, NotApplicable=6, Pending=6, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Non-Compliant | 12\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout configured\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout configured\n- malware.antivirus | Antivirus | Compliant | ESET Endpoint Security (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Non-Compliant | off\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Non-Compliant | block\n- firewall.private.enabled | Private firewall | Non-Compliant | off\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Non-Compliant | block\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | password-protected screen saver on for all signed-in users\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; too long or off for: CONTOSO\\tnguyen (1200s, active=1)\n- screen_lock.password | Screen lock password requirement | Non-Compliant | screen saver does not require sign-in for: CONTOSO\\tnguyen\n- usb.device_access | USB storage device access | Non-Compliant | deny\n- usb.read | USB storage read | Non-Compliant | deny\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | CONTOSO\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-6F4VUQI\\jdoe\n- user.account_type | Signed-in user account type | Compliant | Standard (CONTOSO\\tnguyen)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 1 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 50 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | AUOptions=2\n- gdpr.updates_recent | Security updates installed recently | Compliant | 29 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26200\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: success; account management: none\n- gdpr.time_sync | Time synchronisation | Non-Compliant | type NoSync, service Stopped/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Non-Compliant | enabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | rdp\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | GPT4All\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | GPT4All\n- aiact.risk_classification | AI system risk classification | Pending | GPT4All\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Non-Compliant | Recall feature Enabled, not disabled by policy\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.expiration: Low\n- password.min_age: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- firewall.domain.enabled: High\n- firewall.domain.inbound: High\n- firewall.domain.outbound: Low\n- firewall.private.enabled: High\n- firewall.private.inbound: High\n- firewall.private.outbound: Low\n- firewall.public.inbound: High\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.device_access: Low\n- usb.read: Low\n- admin.unauthorized: High\n- encryption.protection: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.security_logging: Medium\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.ai_register: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=15, assessed=52, percent=29\nTOP GAPS (computed, highest risk first):\n- firewall.domain.enabled | Domain firewall | High | off\n- firewall.domain.inbound | Domain inbound | High | allow\n- firewall.private.enabled | Private firewall | High | off\n- firewall.private.inbound | Private inbound | High | allow\n- firewall.public.inbound | Public inbound | High | allow\nFIX GUIDANCE (highest-risk gaps):\n- firewall.domain.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- firewall.private.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-6F4VUQI - Microsoft Windows 11 Pro 10.0.26200\", \"score\": {\"compliant\": 15, \"assessed\": 52, \"percent\": 29}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 37, \"NotApplicable\": 6, \"Pending\": 6, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.domain.enabled\", \"name\": \"Domain firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.private.enabled\", \"name\": \"Private firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.public.inbound\", \"name\": \"Public inbound\", \"risk\": \"High\", \"found\": \"allow\"}], \"summary\": \"DESKTOP-6F4VUQI meets 15 of 52 assessed checks (29%). The most serious gaps are Domain firewall, Domain inbound and Private firewall. 6 items are pending.\", \"next_steps\": [\"Domain firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Private firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Answer the 4 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall on, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"Firewall inbound policy is firewall on, stealth mode off instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Block or allow external storage with your MDM (Restrictions / device control) or an endpoint DLP product.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 37 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 33; snaps: slack, firefox, chromium; flatpaks: org.mozilla.firefox\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"The device reports 37 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: on\nREQUIRED: off\nDETAIL: A connected computer can read and change data on the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Non-Compliant\", \"title\": \"USB debugging off is non-compliant\", \"finding\": \"USB debugging off is on instead of off. USB debugging lets a connected computer install apps and read data without the usual checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Anyone who can plug the unlocked phone into a computer can copy data or install software.\"}, \"recommendation\": [\"Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; USB debugging should be off.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000367, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period is compliant\", \"finding\": \"Lockout observation period is set to 30 minutes, in line with the required >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: on\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Error\", \"title\": \"Firewall: evidence missing\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, XtsAes128\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker protection\", \"finding\": \"Found: protection Off, XtsAes128. Required: protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rooted or bootloader unlocked\nREQUIRED: not rooted\nDETAIL: Android security protections are weakened on a rooted phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not rooted: does not meet the baseline\", \"finding\": \"Found: rooted or bootloader unlocked. Required: not rooted. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"AI transparency to users is partial instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 265 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 15 applications inventoried, 12 recognised (23 common applications covered). Adobe Acrobat Reader 25.001.20529 (49 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-43550, CVE-2025-43573); Firefox 152.3 (201 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100768, CVE-2026-100779); VLC 3.0.3 (11 CVEs, critical; fixed in 3.0.20; e.g. CVE-2019-12874, CVE-2019-13602); Safari 26.6 (3 CVEs, high; fixed in 26.6.1; e.g. CVE-2026-43794, CVE-2026-65390); LibreOffice 24.2.5 (1 CVE, high; fixed in 24.2.5.1; e.g. CVE-2024-6472). 86 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 265 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: within limit\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: lmuller\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout matches the required setting\", \"finding\": \"Screen lock timeout is set to within limit, in line with the required <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Screen timeout 30 minutes\nREQUIRED: <= 300 seconds\nDETAIL: The phone stays unlocked for 30 minutes when not in use.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen timeout\", \"finding\": \"The user reports Screen timeout 30 minutes, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: auditd not running, journal persistent\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging is non-compliant\", \"finding\": \"Security event logging is auditd not running, journal persistent instead of auditd running. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Android 15\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Chrome 138).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version matches the required setting\", \"finding\": \"Supported operating system version is set to Android 15, in line with the required >= Android 13.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: restic-backup.timer.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: decision pending\", \"finding\": \"Current state: date not readable. Confirm the last backup date in: restic-backup.timer.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (passcode set)\nREQUIRED: on\nDETAIL: iOS encrypts the data on the phone with hardware keys once a passcode is set.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Data protection (encryption) matches the required setting\", \"finding\": \"Data protection (encryption) is set to on (passcode set), in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: mkhan, jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"The device reports mkhan, jdoe, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: required\nREQUIRED: password required\nDETAIL: Users checked: rschmit\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement matches the required setting\", \"finding\": \"Found: required. This satisfies the baseline requirement of password required.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain inbound is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no screen lock (Swipe/None)\nREQUIRED: PIN, password or pattern\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Screen lock is no screen lock (Swipe/None) instead of PIN, password or pattern. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"Lockout observation period is not set instead of >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nDETAIL: pam_pwquality not enabled; pam_unix minimum\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password length\", \"finding\": \"The device reports 6, but the baseline requires >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 600 idle, password delay 60\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: 600 idle, password delay 60. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Unauthorized employee administrator accounts on Linux: Unauthorized employee administrator accounts on Linux. Why it matters: Everyday user accounts with administrator rights let malware and mistakes change anything on the device. Risk if missing: High - Malware running as an administrator can disable protection, install persistence and steal credentials. Requirement: No administrator accounts other than APPROVED_ADMINS (and the disabled built-in Administrator on Windows). How to fix on Linux: (1) Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d. How to verify: Re-run the audit; Unauthorized administrator accounts should read none. Related controls: ISO 27001:2022 A.8.2 (Privileged access rights); ISO 27001:2022 A.5.18 (Access rights); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Signed-in user account type on Linux: Signed-in user account type on Linux. Why it matters: Day-to-day work under an administrator account lets any malware or mistake change the whole system. Risk if missing: High - Malware started by the signed-in user runs with full administrator rights. Requirement: The everyday signed-in user is a Standard account (not an administrator), unless listed in APPROVED_ADMINS. How to fix on Linux: (1) Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account. How to verify: Re-run the audit; Signed-in user account type should read Standard. Related controls: ISO 27001:2022 A.8.2 (Privileged access rights); ISO 27001:2022 A.5.15 (Access control); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Backup configured on Linux: Backup configured on Linux. Why it matters: Backups make it possible to restore personal data after loss, theft, failure or ransomware. Risk if missing: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware. Requirement: A backup solution is configured for user data. How to fix on Linux: (1) Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup). How to verify: Re-run the audit; Backup configured should be Compliant. Related controls: ISO 27001:2022 A.8.13 (Information backup); GDPR Art. 32(1)(c) (Security of processing: restore availability and access after an incident).\n[4] Approved IT administrator on Linux: Approved IT administrator on Linux. Why it matters: A known, approved IT account should manage the device so that support does not depend on users' own admin rights. Risk if missing: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights. Requirement: At least one account listed in APPROVED_ADMINS is an enabled administrator. How to fix on Linux: (1) Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>. How to verify: Re-run the audit; Approved IT administrator should list the approved account. Related controls: ISO 27001:2022 A.8.2 (Privileged access rights); ISO 27001:2022 A.5.15 (Access control); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer, Firefox 143; detected by the browser, security settings not checked.\n\nQUESTION: How can I set up approved it administrator on my Linux?"}, {"role": "assistant", "content": "A known, approved IT account should manage the device so that support does not depend on users' own admin rights. To fix it on Linux [4]:\n1. Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\n\nTo check it worked: Re-run the audit; Approved IT administrator should list the approved account."}], "meta": {"task": "chat", "os": "linux", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: asmith\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"The device reports asmith, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Compliant\", \"title\": \"Public firewall is compliant\", \"finding\": \"Public firewall is set to on, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Sequoia 15.8.3-24H686, Command Line Tools for Xcode-16.4.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"The device reports 2 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"The device reports no lockout, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Tahoe 26.5.3-23J783, Command Line Tools for Xcode-16.4.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates: does not meet the baseline\", \"finding\": \"The device reports 2 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Compliant\", \"title\": \"AI transparency to users is compliant\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 50\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: NTP service yes, synchronised yes\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation matches the required setting\", \"finding\": \"Time synchronisation is set to NTP service yes, synchronised yes, in line with the required time synchronised.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 181 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 19 applications inventoried, 14 recognised (31 common applications covered). Google Chrome 153.0.8010.36 (171 CVEs, critical; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301); Microsoft Edge 151.0.4129.78 (7 CVEs, high; fixed in 153.0.4234.46; e.g. CVE-2026-69486, CVE-2026-70341); Microsoft Teams 8.3.1 (3 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2025-49737, CVE-2025-53783). 172 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"The device reports 181 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"Found: none running. Required: antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification does not apply here\", \"finding\": \"Not applicable on this device: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Breach response procedure is no instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"The device reports USB drives writable without encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: SALES-MacBook-17\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=44, Non-Compliant=4, NotApplicable=4, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5276\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5386, 26 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 125 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 0 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (lmuller)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 15 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | Time Machine\n- gdpr.backup_recent | Recent backup | Non-Compliant | 43 days ago\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | ChatGPT\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- gdpr.backup_recent: Medium\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\nSCORE (computed): compliant=44, assessed=48, percent=92\nTOP GAPS (computed, highest risk first):\n- gdpr.backup_recent | Recent backup | Medium | 43 days ago\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | ChatGPT\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Medium | not restricted\n- org.dpia | Data protection impact assessment | Medium | partial\nFIX GUIDANCE (highest-risk gaps):\n- gdpr.backup_recent: Run tmutil startbackup and check that the destination is reachable.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- aiact.os_ai_features: Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow)."}, {"role": "assistant", "content": "{\"device\": \"SALES-MacBook-17 - macOS 26.0.1\", \"score\": {\"compliant\": 44, \"assessed\": 48, \"percent\": 92}, \"counts\": {\"Compliant\": 44, \"Non-Compliant\": 4, \"NotApplicable\": 4, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"gdpr.backup_recent\", \"name\": \"Recent backup\", \"risk\": \"Medium\", \"found\": \"43 days ago\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"ChatGPT\"}, {\"check_id\": \"aiact.os_ai_features\", \"name\": \"Apple Intelligence external integrations restricted\", \"risk\": \"Medium\", \"found\": \"not restricted\"}, {\"check_id\": \"org.dpia\", \"name\": \"Data protection impact assessment\", \"risk\": \"Medium\", \"found\": \"partial\"}], \"summary\": \"SALES-MacBook-17 meets 44 of 48 assessed checks (92%). There are 4 gaps, none of them rated high risk. 4 items are pending.\", \"next_steps\": [\"Recent backup: Run tmutil startbackup and check that the destination is reachable.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Apple Intelligence external integrations restricted: Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off (no passcode)\nREQUIRED: on\nDETAIL: Without a passcode, iOS data protection is not active.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Data protection (encryption) is non-compliant\", \"finding\": \"The user reports off (no passcode), but the baseline requires on. Full-disk encryption keeps data unreadable if the device is lost or stolen. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 11 Pro build 22621\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"The device reports Microsoft Windows 11 Pro build 22621, but the baseline requires build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone is non-compliant\", \"finding\": \"Find My iPhone is off instead of on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: DeepSeek, Gemini, Replika\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The user reports DeepSeek, Gemini, Replika, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Red Hat Enterprise Linux 8.10 (Ootpa)\nREQUIRED: RHEL 9 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"Supported operating system version is Red Hat Enterprise Linux 8.10 (Ootpa) instead of RHEL 9 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Recall feature Disabled, not disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Windows Recall / Copilot data analysis disabled: does not meet the baseline\", \"finding\": \"Windows Recall / Copilot data analysis disabled is Recall feature Disabled, not disabled by policy instead of disabled by policy. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 14 (android)\nAI TOOLS FOUND: Character.AI, Meta AI, Otter\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Character.AI: AI companion chatbot; provider Character Technologies (US); data: Provider cloud outside the EEA; default risk limited; GDPR: Consumer entertainment chatbot; not suitable for work use. Users must be told they are talking to an AI (AI Act Art. 50).\n- Meta AI: Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps); provider Meta (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\n- Otter: AI meeting transcription; provider Otter.ai (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Records and transcribes other people's voices: needs a lawful basis and notice to meeting participants, plus a data processing agreement.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Character.AI\", \"kind\": \"AI companion chatbot\", \"approved\": false, \"default_risk_class\": \"limited\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer entertainment chatbot; not suitable for work use. Users must be told they are talking to an AI (AI Act Art. 50).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Meta AI\", \"kind\": \"Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Otter\", \"kind\": \"AI meeting transcription\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Records and transcribes other people's voices: needs a lawful basis and notice to meeting participants, plus a data processing agreement.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Character.AI, Meta AI, Otter.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: on\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Error\", \"title\": \"Domain firewall: could not be checked\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Minimum password length is 10 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Analytics sharing minimised\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Compliant\", \"title\": \"Firewall matches the required setting\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 14\nDETAIL: pwpolicy global account policy\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (macos):\n- Deploy a Passcode configuration profile through your MDM with minLength = 14.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Minimum password length is 10 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile through your MDM with minLength = 14.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 20\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold is non-compliant\", \"finding\": \"Account lockout threshold is 20 instead of 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: set automatically\nREQUIRED: set automatically\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Date & time set automatically matches the required setting\", \"finding\": \"Date & time set automatically is set to set automatically, in line with the required set automatically. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted: does not meet the baseline\", \"finding\": \"Found: assistant screen access on. Required: screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 10 (android)\nHOSTNAME: ops-galaxy-06\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=12, Non-Compliant=7, NotApplicable=0, Pending=11, Error=0\nEVIDENCE: 2 detected by the browser, 20 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Non-Compliant | no screen lock (Swipe/None)\n- screen_lock.timeout | Screen timeout | Pending | not answered\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | 4-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 10 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2025-11 (11 months old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 10\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | no notifications on lock screen\n- aiact.inventory | AI software inventory | Compliant | Gemini, Otter\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Gemini, Otter\n- aiact.risk_classification | AI system risk classification | Pending | Gemini, Otter\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- mobile.passcode_strength: Medium\n- vuln.pending_updates: High\n- gdpr.auto_updates: High\n- gdpr.os_supported: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\nSCORE (computed): compliant=12, assessed=19, percent=63\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | Android security update | High | security patch 2025-11 (11 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.os_supported | Supported operating system version | High | Android 10\n- screen_lock.enabled | Screen lock | Medium | no screen lock (Swipe/None)\n- mobile.passcode_strength | Screen lock strength | Medium | 4-digit PIN\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\n- gdpr.auto_updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker).\n- gdpr.os_supported: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone."}, {"role": "assistant", "content": "{\"device\": \"ops-galaxy-06 - Android 10\", \"score\": {\"compliant\": 12, \"assessed\": 19, \"percent\": 63}, \"counts\": {\"Compliant\": 12, \"Non-Compliant\": 7, \"NotApplicable\": 0, \"Pending\": 11, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2025-11 (11 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 10\"}, {\"check_id\": \"screen_lock.enabled\", \"name\": \"Screen lock\", \"risk\": \"Medium\", \"found\": \"no screen lock (Swipe/None)\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Screen lock strength\", \"risk\": \"Medium\", \"found\": \"4-digit PIN\"}], \"summary\": \"ops-galaxy-06 meets 12 of 19 assessed checks (63%). The most serious gaps are Android security update, Automatic system updates and Supported operating system version. 11 items are pending. 20 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Automatic system updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker).\", \"Supported operating system version: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"The device reports disabled, but the baseline requires enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"Data protection impact assessment is partial instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound is non-compliant\", \"finding\": \"Private inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Microsoft Windows 11 Pro build 26100\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version is compliant\", \"finding\": \"The device reports Microsoft Windows 11 Pro build 26100, which meets the requirement (build 22631 or later).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after 300 seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"Found: password after 300 seconds. Required: password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: FileVault off\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Encryption\", \"finding\": \"The device reports FileVault off, but the baseline requires FileVault on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block (firewall on, stealth mode on)\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Compliant\", \"title\": \"Firewall inbound policy is compliant\", \"finding\": \"The device reports block (firewall on, stealth mode on), which meets the requirement (block).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: usb-storage module disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Non-Compliant\", \"title\": \"USB storage device access: does not meet the baseline\", \"finding\": \"The device reports deny, but the baseline requires allow. Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The device access setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage device access should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: automatic updates turned off by policy\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates: does not meet the baseline\", \"finding\": \"The device reports automatic updates turned off by policy, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Compliant\", \"title\": \"BitLocker: meets the baseline\", \"finding\": \"Found: on. This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised is non-compliant\", \"finding\": \"Usage & diagnostics sharing minimised is usage & diagnostics on instead of usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 37 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: python3-libs, glibc, libxml2. Uses the cached repository data.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Found: 37 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iPadOS 27.0 (ios)\nHOSTNAME: MKT-iPad-06\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=16, Non-Compliant=2, NotApplicable=3, Pending=8, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 1 minute\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store and employer-managed apps\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iPadOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iPadOS 27.0\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | backed up to a computer\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- gdpr.telemetry: Low\nSCORE (computed): compliant=16, assessed=18, percent=89\nTOP GAPS (computed, highest risk first):\n- mobile.passcode_strength | Passcode strength | Medium | 4-digit code\n- gdpr.telemetry | Analytics sharing minimised | Low | analytics sharing on\nFIX GUIDANCE (highest-risk gaps):\n- mobile.passcode_strength: Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\n- gdpr.telemetry: Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics."}, {"role": "assistant", "content": "{\"device\": \"MKT-iPad-06 - iPadOS 27.0\", \"score\": {\"compliant\": 16, \"assessed\": 18, \"percent\": 89}, \"counts\": {\"Compliant\": 16, \"Non-Compliant\": 2, \"NotApplicable\": 3, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.passcode_strength\", \"name\": \"Passcode strength\", \"risk\": \"Medium\", \"found\": \"4-digit code\"}, {\"check_id\": \"gdpr.telemetry\", \"name\": \"Analytics sharing minimised\", \"risk\": \"Low\", \"found\": \"analytics sharing on\"}], \"summary\": \"MKT-iPad-06 meets 16 of 18 assessed checks (89%). There are 2 gaps, none of them rated high risk. 8 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Passcode strength: Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\", \"Analytics sharing minimised: Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 15.6 (macos)\nHOSTNAME: SALES-MacBook-39\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=44, Non-Compliant=5, NotApplicable=6, Pending=1, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | not set\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5245\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5358, 15 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 180 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 0 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (pmartin)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 2 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 15.6\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | Time Machine\n- gdpr.backup_recent | Recent backup | Compliant | 1 days ago\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | restricted by profile\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.expiration: Low\n- usb.write: Medium\n- gdpr.removable_encryption: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=44, assessed=49, percent=90\nTOP GAPS (computed, highest risk first):\n- password.min_length | Minimum password length | Medium | not set\n- usb.write | USB storage write | Medium | allow\n- gdpr.removable_encryption | Removable media encryption | Medium | external drives writable without enforced encryption\n- org.human_oversight | Human oversight for high-risk AI | Medium | partial\n- password.expiration | Password expiration | Low | never\nFIX GUIDANCE (highest-risk gaps):\n- password.min_length: Deploy a Passcode configuration profile through your MDM with minLength = 14.\n- usb.write: Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\n- gdpr.removable_encryption: Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives."}, {"role": "assistant", "content": "{\"device\": \"SALES-MacBook-39 - macOS 15.6\", \"score\": {\"compliant\": 44, \"assessed\": 49, \"percent\": 90}, \"counts\": {\"Compliant\": 44, \"Non-Compliant\": 5, \"NotApplicable\": 6, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"password.min_length\", \"name\": \"Minimum password length\", \"risk\": \"Medium\", \"found\": \"not set\"}, {\"check_id\": \"usb.write\", \"name\": \"USB storage write\", \"risk\": \"Medium\", \"found\": \"allow\"}, {\"check_id\": \"gdpr.removable_encryption\", \"name\": \"Removable media encryption\", \"risk\": \"Medium\", \"found\": \"external drives writable without enforced encryption\"}, {\"check_id\": \"org.human_oversight\", \"name\": \"Human oversight for high-risk AI\", \"risk\": \"Medium\", \"found\": \"partial\"}, {\"check_id\": \"password.expiration\", \"name\": \"Password expiration\", \"risk\": \"Low\", \"found\": \"never\"}], \"summary\": \"SALES-MacBook-39 meets 44 of 49 assessed checks (90%). There are 5 gaps, none of them rated high risk. 1 items are pending.\", \"next_steps\": [\"Minimum password length: Deploy a Passcode configuration profile through your MDM with minLength = 14.\", \"USB storage write: Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\", \"Removable media encryption: Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 outdated\nREQUIRED: <= 0 outdated\nDETAIL: VLC media player VideoLAN.VLC 89.6.935 89.7.441 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"Outdated third-party applications is 1 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: OneDrive folder backup (policy).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Confirm the last backup date in: OneDrive folder backup (policy). Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 18.7 (ios)\nHOSTNAME: FIN-iPhone-02\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=9, Non-Compliant=10, NotApplicable=2, Pending=8, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Non-Compliant | jailbroken\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store and employer-managed apps\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 18.7 (update to 18.7.10 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 207 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | download only\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 18.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | ChatGPT extension off\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\nSCORE (computed): compliant=9, assessed=19, percent=47\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not jailbroken | High | jailbroken\n- vuln.pending_updates | iOS security updates | High | iOS 18.7 (update to 18.7.10 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 207 critical/high\n- gdpr.auto_updates | Automatic iOS updates | High | download only\n- screen_lock.timeout | Auto-Lock | Medium | Auto-Lock Never\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"FIN-iPhone-02 - iOS 18.7\", \"score\": {\"compliant\": 9, \"assessed\": 19, \"percent\": 47}, \"counts\": {\"Compliant\": 9, \"Non-Compliant\": 10, \"NotApplicable\": 2, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not jailbroken\", \"risk\": \"High\", \"found\": \"jailbroken\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 18.7 (update to 18.7.10 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"207 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"download only\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Auto-Lock\", \"risk\": \"Medium\", \"found\": \"Auto-Lock Never\"}], \"summary\": \"FIN-iPhone-02 meets 9 of 19 assessed checks (47%). The most serious gaps are Not jailbroken, iOS security updates and Known vulnerabilities in iOS. 8 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not jailbroken: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: set automatically\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Pending\", \"title\": \"Date & time set automatically is pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000367, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: check, download and install automatic\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates: meets the baseline\", \"finding\": \"The device reports check, download and install automatic, which meets the requirement (automatic).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register is non-compliant\", \"finding\": \"Found: no. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"Non-Compliant\", \"title\": \"Tamper protection: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Tamper protection stops malware or users switching off Defender.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Attackers commonly disable antivirus before deploying ransomware.\"}, \"recommendation\": [\"Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Tamper protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: fully encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"Disk encryption protection is compliant\", \"finding\": \"The device reports fully encrypted, which meets the requirement (fully encrypted).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: no unencrypted writes\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Error\", \"title\": \"Removable media encryption: could not be checked\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: ClamAV malware protection service\", \"finding\": \"Found: not running. Required: running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI register\", \"finding\": \"Found: partial. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"AI literacy training is partial instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pattern\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: Pattern locks are easy to guess from screen smudges or by watching.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength: does not meet the baseline\", \"finding\": \"Found: pattern. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 104 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-06 Cumulative Update for Windows 11 (2026-06-20)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"The device reports 104 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, None\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker protection: does not meet the baseline\", \"finding\": \"Found: protection Off, None. Required: protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 600 idle, password delay 5\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Screen lock timeout is 600 idle, password delay 5 instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 300 seconds (machine inactivity limit)\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout matches the required setting\", \"finding\": \"Found: 300 seconds (machine inactivity limit). This satisfies the baseline requirement of <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; too long or off for: CONTOSO\\tnguyen (1200s, active=1)\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: no machine inactivity limit; too long or off for: CONTOSO\\\\tnguyen (1200s, active=1). Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"Found: share Mac analytics=1, share with app developers=0. Required: analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 38 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-08-28 23:40:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security updates installed recently\", \"finding\": \"The device reports 38 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: no machine inactivity limit; screen saver off. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: none enabled\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Compliant\", \"title\": \"Remote access services restricted is compliant\", \"finding\": \"Found: none enabled. This satisfies the baseline requirement of only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: pmartin (3600s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: pmartin, asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"The device reports too long for: pmartin (3600s), but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: auditd running, journal volatile\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging is compliant\", \"finding\": \"The device reports auditd running, journal volatile, which meets the requirement (auditd running).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 16 (android)\nHOSTNAME: mkt-galaxy-40\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=18, Non-Compliant=2, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 5 minutes\n- mobile.passcode_strength | Screen lock strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 16 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-10 (0 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Compliant | Android 16\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | phone maker cloud backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | Gemini\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Gemini\n- aiact.risk_classification | AI system risk classification | Pending | Gemini\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- vuln.outdated_apps: Medium\n- aiact.unapproved: Medium\nSCORE (computed): compliant=18, assessed=20, percent=90\nTOP GAPS (computed, highest risk first):\n- vuln.outdated_apps | App updates | Medium | automatic app updates off\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | Gemini\nFIX GUIDANCE (highest-risk gaps):\n- vuln.outdated_apps: Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it."}, {"role": "assistant", "content": "{\"device\": \"mkt-galaxy-40 - Android 16\", \"score\": {\"compliant\": 18, \"assessed\": 20, \"percent\": 90}, \"counts\": {\"Compliant\": 18, \"Non-Compliant\": 2, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.outdated_apps\", \"name\": \"App updates\", \"risk\": \"Medium\", \"found\": \"automatic app updates off\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"Gemini\"}], \"summary\": \"mkt-galaxy-40 meets 18 of 20 assessed checks (90%). There are 2 gaps, none of them rated high risk. 10 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"App updates: Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound is non-compliant\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall: does not meet the baseline\", \"finding\": \"The device reports no active firewall, but the baseline requires active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: AllowTelemetry=3 (Settings)\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"The device reports AllowTelemetry=3 (Settings), but the baseline requires required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall outbound policy is non-compliant\", \"finding\": \"Firewall outbound policy is no firewall instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"The device reports not encrypted, but the baseline requires recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: all apps: Not allowed\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer).\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Pending\", \"title\": \"Install unknown apps blocked: decision pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: restricted by profile\nREQUIRED: restricted by profile\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Compliant\", \"title\": \"Apple Intelligence external integrations restricted is compliant\", \"finding\": \"Found: restricted by profile. This satisfies the baseline requirement of restricted by profile.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: lmuller, mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee administrator accounts\", \"finding\": \"The device reports lmuller, mkhan, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 15 (android)\nHOSTNAME: dev-galaxy-09\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=9, Non-Compliant=11, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 5 minutes\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | 4-digit PIN\n- mobile.jailbreak_root | Not rooted | Pending | not answered\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Pending | not answered\n- encryption.enabled | Device encryption | Compliant | on (Android 15 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2026-05 (5 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Compliant | Android 15\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | Grok\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Grok\n- aiact.risk_classification | AI system risk classification | Pending | Grok\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\n- org.processor_agreements: High\nSCORE (computed): compliant=9, assessed=20, percent=45\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- vuln.pending_updates | Android security update | High | security patch 2026-05 (5 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.backup_configured | Google backup | High | no backup\n- org.processor_agreements | Data processing agreements | High | no\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\n- gdpr.auto_updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker)."}, {"role": "assistant", "content": "{\"device\": \"dev-galaxy-09 - Android 15\", \"score\": {\"compliant\": 9, \"assessed\": 20, \"percent\": 45}, \"counts\": {\"Compliant\": 9, \"Non-Compliant\": 11, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2026-05 (5 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Google backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"org.processor_agreements\", \"name\": \"Data processing agreements\", \"risk\": \"High\", \"found\": \"no\"}], \"summary\": \"dev-galaxy-09 meets 9 of 20 assessed checks (45%). The most serious gaps are Install unknown apps blocked, Android security update and Automatic system updates. 10 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Automatic system updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker).\", \"Answer the 6 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: macOS 26.0.1\nREQUIRED: macOS 14 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (macos):\n- Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version: meets the baseline\", \"finding\": \"Supported operating system version is set to macOS 26.0.1, in line with the required macOS 14 or later.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2026-06 (4 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2026-06 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Android security update is non-compliant\", \"finding\": \"Android security update is security patch 2026-06 (4 months old) instead of <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: set automatically\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Pending\", \"title\": \"Date & time set automatically: decision pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training: meets the baseline\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: Claude. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption protection\", \"finding\": \"Found: not encrypted. Required: active, strong cipher. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 116 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-06-11  05:59:42\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Found: 116 days ago. Required: <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption protection: does not meet the baseline\", \"finding\": \"Found: not encrypted. Required: fully encrypted. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: screen saver within limit for: FABRIKAM\\sali\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout: meets the baseline\", \"finding\": \"Screen lock timeout is set to screen saver within limit for: FABRIKAM\\\\sali, in line with the required <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iCloud Backup\", \"finding\": \"iCloud Backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"Found: not enforced. Required: enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: logon: success+failure; account management: success+failure\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging: meets the baseline\", \"finding\": \"The device reports logon: success+failure; account management: success+failure, which meets the requirement (logon success+failure and account changes audited).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: mkhan\nREQUIRED: password required\nDETAIL: Users checked: mkhan\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"Screen lock password requirement is not required for: mkhan instead of password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification does not apply here\", \"finding\": \"Not applicable on this device: no AI tools found. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 365 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration is compliant\", \"finding\": \"The device reports 365 days, which meets the requirement (1-365 days).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 69 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 757 packages checked against rhel:8 advisories. mod_ssl-debuginfo 1:2.4.37-39.module+el8.4.0+12865+a7065a39.1 (60 CVEs, critical, actively exploited; fixed in 1:2.4.37-65.module+el8.10.0+24281+ea10630c.8; e.g. CVE-2006-20001, CVE-2022-36760); glibc-langpack-or 0:2.28-225.el8_8.6 (6 CVEs, high; fixed in 0:2.28-251.el8_10.22; e.g. CVE-2024-2961, CVE-2024-33599); pam 0:1.3.1-37.el8_10 (2 CVEs, high; fixed in 0:1.3.1-38.el8_10; e.g. CVE-2025-6020, CVE-2025-8941); subversion-devel 0:1.10.2-4.module+el8.3.0+9886+ac338b6d (1 CVE, high; fixed in 0:1.10.2-5.module+el8.6.0+15157+188c9801; e.g. CVE-2022-24070). 16 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Found: 69 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 16\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Compliant\", \"title\": \"Minimum password length is compliant\", \"finding\": \"Minimum password length is set to 16, in line with the required >= 14.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is set to none, in line with the required only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Found: no. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Signed-in user account type on macOS: Signed-in user account type on macOS. Why it matters: Day-to-day work under an administrator account lets any malware or mistake change the whole system. Risk if missing: High - Malware started by the signed-in user runs with full administrator rights. Requirement: The everyday signed-in user is a Standard account (not an administrator), unless listed in APPROVED_ADMINS. How to fix on macOS: (1) Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account. How to verify: Re-run the audit; Signed-in user account type should read Standard. Related controls: ISO 27001:2022 A.8.2 (Privileged access rights); ISO 27001:2022 A.5.15 (Access control); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] How to install on Windows: Run the downloaded setup .exe. Windows SmartScreen shows 'Windows protected your PC' because the app is not yet code-signed: click More info, then Run anyway. Click Next, Install and Finish. When the audit starts, click Yes in the User Account Control prompt. The Windows version is a beta.\n[3] Account lockout threshold on macOS: Account lockout threshold on macOS. Why it matters: Locking an account after a few wrong passwords stops online guessing. Risk if missing: Medium - Unlimited attempts allow brute-force attacks against local accounts. Requirement: Lockout after 1..LOCKOUT_THRESHOLD_MAX failed attempts (0 = never locks). How to fix on macOS: (1) Deploy a Passcode configuration profile with maxFailedAttempts = 5. How to verify: Re-run the audit; Account lockout threshold should be between 1 and 5. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Screen lock timeout on macOS: Screen lock timeout on macOS. Why it matters: The shorter the idle time before locking, the smaller the window for misuse. Risk if missing: Medium - A long or missing timeout leaves unattended devices open. Requirement: Idle time before lock (including any grace delay) is 1..SCREEN_LOCK_MAX_SECONDS seconds (baseline: 300 = 5 minutes). Phones: Auto-Lock / screen timeout. How to fix on macOS: (1) Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile. How to verify: Re-run the audit; Screen lock timeout should be 300 seconds or less. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: macOS 14.7.8 (macos), audited 2026-10-05; 39 of 49 assessed checks compliant; 10 gaps, 2 pending, 0 not read. Gaps: Account lockout threshold: 10; Lockout observation period: 10 minutes; Screen lock timeout: 1200 idle, password delay 0; Disk encryption recovery: no recovery key; Known vulnerabilities in installed software: 452 critical/high; Backup configured: none; Recent backup: no backup; Remote access services restricted: screensharing, chromeremotedesktop.\n\nQUESTION: Where do I start?"}, {"role": "assistant", "content": "Your audit found 10 gaps. Start with Account lockout threshold (found: 10). Locking an account after a few wrong passwords stops online guessing. To fix it [3]:\n1. Deploy a Passcode configuration profile with maxFailedAttempts = 5.\n\nThen look at: Lockout observation period, Screen lock timeout, Disk encryption recovery."}], "meta": {"task": "chat", "os": "macos", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no signature database\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures: does not meet the baseline\", \"finding\": \"Found: no signature database. Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker\", \"finding\": \"The device reports off, but the baseline requires on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"Found: none of the approved accounts is an administrator. Required: itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted: does not meet the baseline\", \"finding\": \"The user reports assistant screen access on, but the baseline requires screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: GPT4All\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: GPT4All.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: network time on\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Error\", \"title\": \"Time synchronisation: could not be checked\", \"finding\": \"This check returned an error: Requires root: run the collector with sudo. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"The device reports external drives writable without enforced encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] What the Devseis Endpoint Auditor is: The Devseis Endpoint Auditor is a free, open-source, read-only check of a computer or phone against ISO 27001, GDPR technical controls and the EU AI Act. It reads security settings, compares them with a baseline, matches installed software against known vulnerabilities, and an AI auditor that runs on the device writes a report with findings, risk levels and fix steps. It is not a certification and not legal advice.\n[2] Who is Arx?: Arx is the name of the Devseis security assistant and endpoint auditor (from the Latin for fortress). Arx answers questions about securing computers and phones, ISO 27001, GDPR and the EU AI Act, and audits devices: computers with the Devseis Endpoint Auditor desktop app, phones with the phone check in the browser. Arx runs on the device; nothing is uploaded.\n[3] The Devseis AI auditor model: The assistant and the report writer are a small language model fine-tuned by Devseis from Qwen2.5 and run on the device with WebLLM. Audit verdicts are decided by rules in code, not by the model; the model explains, prioritises and writes, and every report answer is checked against the evidence before it is used.\n[4] How to install on a Mac: Open the downloaded .dmg file and drag Devseis Endpoint Auditor to Applications. The first time, macOS says it could not verify the app because it is not yet notarised: click Done, then open System Settings, Privacy & Security, and click Open Anyway. After that it opens normally. It needs macOS 13 or newer.\n\nMY AUDIT: none\n\nMY DEVICE: iOS 18.6 (ios) on an iPhone, Chrome 141 (iOS); detected by the browser, security settings not checked.\n\nQUESTION: What is the Devseis Endpoint Auditor?"}, {"role": "assistant", "content": "The Devseis Endpoint Auditor is a free, open-source, read-only check of a computer or phone against ISO 27001, GDPR technical controls and the EU AI Act. It reads security settings, compares them with a baseline, matches installed software against known vulnerabilities, and an AI auditor that runs on the device writes a report with findings, risk levels and fix steps. It is not a certification and not legal advice. [1]"}], "meta": {"task": "chat", "os": "any", "status": "faq", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: akaya, asmith\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: akaya, asmith. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ConfigDataInstall=0, CriticalUpdateInstall=1\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware security updates: does not meet the baseline\", \"finding\": \"Found: ConfigDataInstall=0, CriticalUpdateInstall=1. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: assistant screen access off\nREQUIRED: screen access off unless approved\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Compliant\", \"title\": \"AI assistant access restricted: meets the baseline\", \"finding\": \"The user reports assistant screen access off, which meets the requirement (screen access off unless approved). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in ArqAgent.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Current state: date not readable. Confirm the last backup date in ArqAgent.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nDETAIL: Users checked: lmuller\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock is compliant\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Ollama, Claude\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: Ollama, Claude. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: enabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Guest account disabled\", \"finding\": \"Guest account disabled is enabled instead of disabled. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"The device reports USB storage writable without enforced encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 16 (android)\nHOSTNAME: ops-xiaomi-32\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=25, Non-Compliant=2, NotApplicable=2, Pending=1, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 30 seconds\n- mobile.passcode_strength | Screen lock strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Non-Compliant | on\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 16 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-08 (2 months old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Compliant | Android 16\n- gdpr.time_sync | Date & time set automatically | Pending | not answered\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- mobile.usb_debugging: Medium\n- gdpr.lock_screen_previews: Low\nSCORE (computed): compliant=25, assessed=27, percent=93\nTOP GAPS (computed, highest risk first):\n- mobile.usb_debugging | USB debugging off | Medium | on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Low | all content shown\nFIX GUIDANCE (highest-risk gaps):\n- mobile.usb_debugging: Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\n- gdpr.lock_screen_previews: Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications."}, {"role": "assistant", "content": "{\"device\": \"ops-xiaomi-32 - Android 16\", \"score\": {\"compliant\": 25, \"assessed\": 27, \"percent\": 93}, \"counts\": {\"Compliant\": 25, \"Non-Compliant\": 2, \"NotApplicable\": 2, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.usb_debugging\", \"name\": \"USB debugging off\", \"risk\": \"Medium\", \"found\": \"on\"}, {\"check_id\": \"gdpr.lock_screen_previews\", \"name\": \"Sensitive notifications hidden when locked\", \"risk\": \"Low\", \"found\": \"all content shown\"}], \"summary\": \"ops-xiaomi-32 meets 25 of 27 assessed checks (93%). There are 2 gaps, none of them rated high risk. 1 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"USB debugging off: Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\", \"Sensitive notifications hidden when locked: Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000367, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_pwquality not enabled\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Password complexity is pam_pwquality not enabled instead of 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: The macOS application firewall filters inbound connections only.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (macos):\n- The macOS application firewall does not filter outbound traffic; use a network filter product if outbound blocking is required.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Compliant\", \"title\": \"Firewall outbound policy is compliant\", \"finding\": \"Firewall outbound policy is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Data processing agreements is no instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"Diagnostic data minimised is not configured instead of required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1200 idle, password delay 60\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"The device reports 1200 idle, password delay 60, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: on\nREQUIRED: off\nDETAIL: A connected computer can read and change data on the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Non-Compliant\", \"title\": \"USB debugging off is non-compliant\", \"finding\": \"USB debugging off is on instead of off. USB debugging lets a connected computer install apps and read data without the usual checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Anyone who can plug the unlocked phone into a computer can copy data or install software.\"}, \"recommendation\": [\"Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; USB debugging should be off.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answer ORG_ROPA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Password complexity is not enforced instead of enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MIN_DAYS 0\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (linux):\n- Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"Found: PASS_MIN_DAYS 0. Required: >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted is non-compliant\", \"finding\": \"The device reports not restricted, but the baseline requires restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: not rooted (user confirmed)\nREQUIRED: not rooted\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Compliant\", \"title\": \"Not rooted is compliant\", \"finding\": \"Not rooted is set to not rooted (user confirmed), in line with the required not rooted. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Password complexity is not enforced instead of enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Operating system AI features restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: none built in\nREQUIRED: restricted by policy\nDETAIL: Linux distributions ship no built-in AI assistant.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (linux):\n- No action needed: Linux distributions have no built-in AI assistant.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Operating system AI features restricted: not applicable\", \"finding\": \"Not applicable on this device: Linux distributions ship no built-in AI assistant.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Found: partial. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"AI literacy training is no instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Found: no. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically: does not meet the baseline\", \"finding\": \"The user reports set by hand, but the baseline requires set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › General › Date & Time: turn on Set Automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"The device reports no active firewall, but the baseline requires active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (lmuller)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"Found: Administrator (lmuller). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"Breach response procedure is partial instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nHOSTNAME: DESKTOP-J0Q0Y2I\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=13, Non-Compliant=40, NotApplicable=4, Pending=5, Error=2\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 0\n- password.history | Password history | Non-Compliant | 5\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout configured\n- lockout.observation | Lockout observation period | Error | not read\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Non-Compliant | service True, antivirus False\n- malware.realtime | Real-time protection | Non-Compliant | off\n- malware.on_access | On-access protection | Non-Compliant | off\n- malware.tamper | Tamper protection | Non-Compliant | off\n- malware.signatures | Security intelligence | Non-Compliant | 19 days old (version 1.434.479.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | allow\n- screen_lock.enabled | Screen lock | Non-Compliant | no machine inactivity limit; no signed-in user to check\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | screen saver does not require sign-in for: DESKTOP-J0Q0Y2I\\ewagner\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an enabled administrator\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-J0Q0Y2I\\tnguyen\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (DESKTOP-J0Q0Y2I\\ewagner)\n- encryption.feature | BitLocker | Non-Compliant | off\n- encryption.enabled | Encryption | Non-Compliant | EncryptionInProgress (42%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, None\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Error | not read\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 62 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | automatic updates turned off by policy\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 90 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 11 Pro build 22621\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: none; account management: none\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Non-Compliant | enabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | rdp\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.expiration: Low\n- password.min_age: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- malware.service: High\n- malware.realtime: High\n- malware.on_access: Medium\n- malware.tamper: Medium\n- malware.signatures: High\n- firewall.private.inbound: High\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.feature: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.os_supported: High\n- gdpr.security_logging: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.dpia: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=13, assessed=53, percent=25\nTOP GAPS (computed, highest risk first):\n- malware.service | Microsoft Defender Antimalware | High | service True, antivirus False\n- malware.realtime | Real-time protection | High | off\n- malware.signatures | Security intelligence | High | 19 days old (version 1.434.479.0)\n- firewall.private.inbound | Private inbound | High | allow\n- firewall.public.enabled | Public firewall | High | off\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\n- malware.realtime: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\n- malware.signatures: Run Update-MpSignature and check that the device can reach Windows Update or your update server."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-J0Q0Y2I - Microsoft Windows 11 Pro 10.0.22621\", \"score\": {\"compliant\": 13, \"assessed\": 53, \"percent\": 25}, \"counts\": {\"Compliant\": 13, \"Non-Compliant\": 40, \"NotApplicable\": 4, \"Pending\": 5, \"Error\": 2}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"Microsoft Defender Antimalware\", \"risk\": \"High\", \"found\": \"service True, antivirus False\"}, {\"check_id\": \"malware.realtime\", \"name\": \"Real-time protection\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Security intelligence\", \"risk\": \"High\", \"found\": \"19 days old (version 1.434.479.0)\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.public.enabled\", \"name\": \"Public firewall\", \"risk\": \"High\", \"found\": \"off\"}], \"summary\": \"DESKTOP-J0Q0Y2I meets 13 of 53 assessed checks (25%). The most serious gaps are Microsoft Defender Antimalware, Real-time protection and Security intelligence. 2 checks could not be read. 5 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Microsoft Defender Antimalware: Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\", \"Real-time protection: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\", \"Security intelligence: Run Update-MpSignature and check that the device can reach Windows Update or your update server.\", \"Answer the 5 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"Found: partial. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: >= 15 minutes\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Error\", \"title\": \"Account lockout duration: evidence missing\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Notification previews hidden when locked\", \"finding\": \"The user reports Always, but the baseline requires When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 14 (android)\nHOSTNAME: sales-pixel-01\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=9, Non-Compliant=10, NotApplicable=0, Pending=11, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Pending | not answered\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 1 minute\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | swipe or none\n- mobile.jailbreak_root | Not rooted | Non-Compliant | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 14 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2026-01 (9 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Compliant | Android 14\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | Character.AI, ChatGPT, Gemini\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Character.AI, ChatGPT, Gemini\n- aiact.risk_classification | AI system risk classification | Pending | Character.AI, ChatGPT, Gemini\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=9, assessed=19, percent=47\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not rooted | High | rooted or bootloader unlocked\n- vuln.pending_updates | Android security update | High | security patch 2026-01 (9 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.backup_configured | Google backup | High | no backup\n- mobile.passcode_strength | Screen lock strength | Medium | swipe or none\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\n- gdpr.auto_updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker)."}, {"role": "assistant", "content": "{\"device\": \"sales-pixel-01 - Android 14\", \"score\": {\"compliant\": 9, \"assessed\": 19, \"percent\": 47}, \"counts\": {\"Compliant\": 9, \"Non-Compliant\": 10, \"NotApplicable\": 0, \"Pending\": 11, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not rooted\", \"risk\": \"High\", \"found\": \"rooted or bootloader unlocked\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2026-01 (9 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Google backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Screen lock strength\", \"risk\": \"Medium\", \"found\": \"swipe or none\"}], \"summary\": \"sales-pixel-01 meets 9 of 19 assessed checks (47%). The most serious gaps are Not rooted, Android security update and Automatic system updates. 11 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not rooted: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Automatic system updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker).\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude, Meta AI, Replika\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Unapproved AI tools (shadow AI) is Claude, Meta AI, Replika instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"Account lockout threshold is 10 instead of 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\", \"Set deny = 5 in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apps only from the App Store\", \"finding\": \"Apps only from the App Store is apps from outside the App Store instead of App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Answer ORG_ROPA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Network Time: Off\nREQUIRED: network time on\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"The device reports Network Time: Off, but the baseline requires network time on. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"sudo systemsetup -setusingnetworktime on.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities is non-compliant\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates: does not meet the baseline\", \"finding\": \"The user reports manual, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nDETAIL: pam_pwquality not enabled; pam_unix minimum\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Found: 6. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: set automatically\nREQUIRED: set automatically\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Date & time set automatically is compliant\", \"finding\": \"Found: set automatically. This satisfies the baseline requirement of set automatically. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: not jailbroken (user confirmed)\nREQUIRED: not jailbroken\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Compliant\", \"title\": \"Not jailbroken: meets the baseline\", \"finding\": \"The user reports not jailbroken (user confirmed), which meets the requirement (not jailbroken). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"Found: anydesk. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (lmuller)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"Found: Administrator (lmuller). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"The device reports none, which meets the requirement (only: none).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"AI register is partial instead of yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: smb\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"Found: smb. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Jan, Hugging Face models, Cursor\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: Jan, Hugging Face models, Cursor. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Answer ORG_AI_TRANSPARENCY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"AI transparency to users is partial instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: popularity-contest on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"Diagnostic data minimised is popularity-contest on instead of usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"Found: off. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall on, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"The device reports firewall on, stealth mode off, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: screensharing\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"The device reports screensharing, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: minclass=0, required classes=4\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Compliant\", \"title\": \"Password complexity is compliant\", \"finding\": \"The device reports minclass=0, required classes=4, which meets the requirement (3 or more character classes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock: does not meet the baseline\", \"finding\": \"The user reports Auto-Lock Never, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: security responses and system data files install automatically\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Compliant\", \"title\": \"Malware security updates: meets the baseline\", \"finding\": \"Malware security updates is set to security responses and system data files install automatically, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 26.7 (ios)\nAI TOOLS FOUND: ChatGPT\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Compliant\", \"title\": \"Data protection impact assessment is compliant\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: lmuller\nREQUIRED: password required\nDETAIL: Users checked: lmuller\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"The device reports not required for: lmuller, but the baseline requires password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nHOSTNAME: DESKTOP-EXYV7YY\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=30, Non-Compliant=22, NotApplicable=9, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Non-Compliant | 5\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 0\n- lockout.duration | Account lockout duration | Non-Compliant | 5 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | ESET Endpoint Security (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Compliant | 180 seconds (machine inactivity limit)\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Non-Compliant | deny\n- usb.read | USB storage read | Non-Compliant | deny\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an enabled administrator\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-EXYV7YY\\sali, DESKTOP-EXYV7YY\\Administrator\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (CONTOSO\\akaya)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 2 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 16 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 22631\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | File History (jdoe)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | winrm\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=0 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.expiration: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- firewall.private.inbound: High\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- usb.device_access: Low\n- usb.read: Low\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.remote_access: High\n- org.ropa: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.human_oversight: Medium\nSCORE (computed): compliant=30, assessed=52, percent=58\nTOP GAPS (computed, highest risk first):\n- firewall.private.inbound | Private inbound | High | allow\n- firewall.public.enabled | Public firewall | High | off\n- firewall.public.inbound | Public inbound | High | allow\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | DESKTOP-EXYV7YY\\sali, DESKTOP-EXYV7YY\\Administrator\n- user.account_type | Signed-in user account type | High | Administrator (CONTOSO\\akaya)\nFIX GUIDANCE (highest-risk gaps):\n- firewall.private.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- firewall.public.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.public.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-EXYV7YY - Microsoft Windows 11 Pro 10.0.22631\", \"score\": {\"compliant\": 30, \"assessed\": 52, \"percent\": 58}, \"counts\": {\"Compliant\": 30, \"Non-Compliant\": 22, \"NotApplicable\": 9, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.public.enabled\", \"name\": \"Public firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.public.inbound\", \"name\": \"Public inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"DESKTOP-EXYV7YY\\\\sali, DESKTOP-EXYV7YY\\\\Administrator\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (CONTOSO\\\\akaya)\"}], \"summary\": \"DESKTOP-EXYV7YY meets 30 of 52 assessed checks (58%). The most serious gaps are Private inbound, Public firewall and Public inbound. 3 items are pending.\", \"next_steps\": [\"Private inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Public firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Public inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Found: not enforced. Required: enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: enabled\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Error\", \"title\": \"Password complexity could not be read\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain outbound is non-compliant\", \"finding\": \"Found: block. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: swipe or none\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: There is no secret needed to unlock the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength: does not meet the baseline\", \"finding\": \"The user reports swipe or none, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Who is Arx?: Arx is the name of the Devseis security assistant and endpoint auditor (from the Latin for fortress). Arx answers questions about securing computers and phones, ISO 27001, GDPR and the EU AI Act, and audits devices: computers with the Devseis Endpoint Auditor desktop app, phones with the phone check in the browser. Arx runs on the device; nothing is uploaded.\n[2] Where to find Android security update on Android: Android security update on Android: What is the Android security update date? Where to look: Settings › About phone › Android version › Android security update (or Settings › Security & privacy › Updates)\n[3] How the phone check works: Phones are checked in the browser: open the Devseis page on the iPhone or Android phone and start the phone check. The browser detects what it can (system version, whether a screen lock is set up) and asks the rest with help on where to look in Settings. Every result is labelled detected or self-reported. The report can be saved or shared as a PDF.\n[4] How to run an audit of a computer: A web page cannot read a computer's settings, so computers are audited with the free Devseis desktop app for Mac, Windows (beta) and Linux. Download it from the Download page, install it, click Run audit and enter the computer password once. The first run downloads the AI auditor; the audit then takes about 5 to 15 minutes and ends with a PDF report.\n\nMY AUDIT: none\n\nMY DEVICE: Android 16 (android) on a Pixel 9a, Chrome 141; detected by the browser, security settings not checked.\n\nQUESTION: What do you know about my computer?"}, {"role": "assistant", "content": "This browser reports Android 16 on a Pixel 9a, using Chrome 141. That is all a web page can see: it can't read security settings such as encryption, the firewall or the screen lock. To check those, run the phone check: open this page on the phone and tap Start the check [3]."}], "meta": {"task": "chat", "os": "android", "status": "device", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain outbound is non-compliant\", \"finding\": \"Domain outbound is block instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: >= 15 minutes\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Error\", \"title\": \"Lockout observation period: evidence missing\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 14.7.8 (macos)\nHOSTNAME: DEV-MacBook-31\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=45, Non-Compliant=2, NotApplicable=7, Pending=1, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5265\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5305, 21 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 120 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (pmartin)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Error | not read\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 38 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 27 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 14.7.8\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | bzserv\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Intel Mac\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- vuln.known_vulnerabilities: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=45, assessed=47, percent=96\nTOP GAPS (computed, highest risk first):\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 38 critical/high\n- org.ai_literacy | AI literacy training | Medium | partial\nFIX GUIDANCE (highest-risk gaps):\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- org.ai_literacy: Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records."}, {"role": "assistant", "content": "{\"device\": \"DEV-MacBook-31 - macOS 14.7.8\", \"score\": {\"compliant\": 45, \"assessed\": 47, \"percent\": 96}, \"counts\": {\"Compliant\": 45, \"Non-Compliant\": 2, \"NotApplicable\": 7, \"Pending\": 1, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"38 critical/high\"}, {\"check_id\": \"org.ai_literacy\", \"name\": \"AI literacy training\", \"risk\": \"Medium\", \"found\": \"partial\"}], \"summary\": \"DEV-MacBook-31 meets 45 of 47 assessed checks (96%). The most serious gaps are Known vulnerabilities in installed software. 1 checks could not be read. 1 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"AI literacy training: Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Windows Defender (on, up to date)\nREQUIRED: enabled and up to date\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (windows):\n- Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus: meets the baseline\", \"finding\": \"The device reports Windows Defender (on, up to date), which meets the requirement (enabled and up to date).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Domain firewall: does not meet the baseline\", \"finding\": \"Domain firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"The device reports allow (no firewall), but the baseline requires deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: >= 1 days\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Error\", \"title\": \"Minimum password age could not be read\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: chromeremotedesktop\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is chromeremotedesktop instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 83 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Recent backup is 83 days ago instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Gemini\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Current state: Gemini. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Private inbound\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LAPTOP-Z2TQAKZ\\jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts is non-compliant\", \"finding\": \"Found: LAPTOP-Z2TQAKZ\\\\jdoe. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"Found: share Mac analytics=1, share with app developers=0. Required: analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (linux):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory is compliant\", \"finding\": \"Found: no AI tools found. This satisfies the baseline requirement of inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates is non-compliant\", \"finding\": \"App updates is automatic app updates off instead of automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answer ORG_AI_TRANSPARENCY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: AUOptions=2\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates: does not meet the baseline\", \"finding\": \"Found: AUOptions=2. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iOS 18.7\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Chrome 138 (iOS)).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version is compliant\", \"finding\": \"Supported operating system version is set to iOS 18.7, in line with the required >= iOS 18.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock strength\", \"finding\": \"Screen lock strength is 4-digit PIN instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: <= 0 pending\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Error\", \"title\": \"Pending security updates: could not be checked\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Ollama, Perplexity, Jan\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Unapproved AI tools (shadow AI) is Ollama, Perplexity, Jan instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ewagner, tnguyen\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee administrator accounts\", \"finding\": \"Found: ewagner, tnguyen. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history is compliant\", \"finding\": \"Found: 24. This satisfies the baseline requirement of >= 24.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google backup\", \"finding\": \"The user reports no backup, but the baseline requires backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: FileVault on\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Encryption matches the required setting\", \"finding\": \"Found: FileVault on. This satisfies the baseline requirement of FileVault on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 27.0 (ios)\nHOSTNAME: LEGAL-iPhone-10\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=16, Non-Compliant=9, NotApplicable=2, Pending=2, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 5 minutes\n- mobile.passcode_strength | Passcode strength | Compliant | 7+ digit numeric code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store only\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | download only\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 27.0\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- mobile.find_my_device: Medium\n- gdpr.auto_updates: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.os_ai_features: Medium\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=16, assessed=25, percent=64\nTOP GAPS (computed, highest risk first):\n- gdpr.auto_updates | Automatic iOS updates | High | download only\n- org.processor_agreements | Data processing agreements | High | no\n- mobile.find_my_device | Find My iPhone | Medium | off\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Medium | ChatGPT extension on\n- org.ai_register | AI register | Medium | no\nFIX GUIDANCE (highest-risk gaps):\n- gdpr.auto_updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\n- org.processor_agreements: List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\n- mobile.find_my_device: Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location."}, {"role": "assistant", "content": "{\"device\": \"LEGAL-iPhone-10 - iOS 27.0\", \"score\": {\"compliant\": 16, \"assessed\": 25, \"percent\": 64}, \"counts\": {\"Compliant\": 16, \"Non-Compliant\": 9, \"NotApplicable\": 2, \"Pending\": 2, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"download only\"}, {\"check_id\": \"org.processor_agreements\", \"name\": \"Data processing agreements\", \"risk\": \"High\", \"found\": \"no\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My iPhone\", \"risk\": \"Medium\", \"found\": \"off\"}, {\"check_id\": \"aiact.os_ai_features\", \"name\": \"Apple Intelligence external integrations restricted\", \"risk\": \"Medium\", \"found\": \"ChatGPT extension on\"}, {\"check_id\": \"org.ai_register\", \"name\": \"AI register\", \"risk\": \"Medium\", \"found\": \"no\"}], \"summary\": \"LEGAL-iPhone-10 meets 16 of 25 assessed checks (64%). The most serious gaps are Automatic iOS updates and Data processing agreements. 2 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Automatic iOS updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\", \"Data processing agreements: List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\", \"Find My iPhone: Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"Compliant\", \"title\": \"Tamper protection: meets the baseline\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Meta AI\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Found: ChatGPT, Meta AI. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"The device reports not configured, but the baseline requires required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Google backup is non-compliant\", \"finding\": \"The user reports no backup, but the baseline requires backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Recall disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=1; AllowRecallEnablement=0; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Compliant\", \"title\": \"Windows Recall / Copilot data analysis disabled matches the required setting\", \"finding\": \"The device reports Recall disabled by policy, which meets the requirement (disabled by policy).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: ClamAV (clamav-daemon)\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus matches the required setting\", \"finding\": \"The device reports ClamAV (clamav-daemon), which meets the requirement (antivirus running).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: security responses and system data files install automatically\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Compliant\", \"title\": \"Malware security updates: meets the baseline\", \"finding\": \"The device reports security responses and system data files install automatically, which meets the requirement (automatic).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: LM Studio\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"The device reports LM Studio, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google Play Protect\", \"finding\": \"Found: Play Protect scanning off. Required: on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Character.AI, ChatGPT, Gemini\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is Character.AI, ChatGPT, Gemini instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 8\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password length\", \"finding\": \"Minimum password length is 8 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: check, download and install automatic\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates is compliant\", \"finding\": \"Found: check, download and install automatic. This satisfies the baseline requirement of automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 16 (android)\nAI TOOLS FOUND: ChatGPT, Claude\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Claude.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google Play Protect\", \"finding\": \"The user reports Play Protect scanning off, but the baseline requires on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted: does not meet the baseline\", \"finding\": \"Found: assistant screen access on. Required: screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"Found: 1 key slots, 0 unlock tokens. Required: recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: swipe or none\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: There is no secret needed to unlock the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock strength\", \"finding\": \"Found: swipe or none. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Found: partial. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Compliant\", \"title\": \"Unauthorized employee administrator accounts is compliant\", \"finding\": \"Found: none. This satisfies the baseline requirement of none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: dnf-automatic timer off, apply_updates=no\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic security updates\", \"finding\": \"Found: dnf-automatic timer off, apply_updates=no. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: macOS 13.7.8\nREQUIRED: macOS 14 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (macos):\n- Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"The device reports macOS 13.7.8, but the baseline requires macOS 14 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Google Chrome Google.Chrome 93.6.698 93.5.2 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Outdated third-party applications is 1 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Compliant\", \"title\": \"Firewall inbound policy matches the required setting\", \"finding\": \"Found: deny. This satisfies the baseline requirement of deny.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training matches the required setting\", \"finding\": \"AI literacy training is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nHOSTNAME: LAPTOP-Y6NRCJU\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=26, Non-Compliant=27, NotApplicable=8, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 0\n- password.history | Password history | Non-Compliant | 0\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Non-Compliant | 5 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | Sophos Endpoint (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Non-Compliant | off\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Non-Compliant | allow\n- firewall.private.enabled | Private firewall | Non-Compliant | off\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Non-Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | block\n- screen_lock.enabled | Screen lock | Compliant | password-protected screen saver on for all signed-in users\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; no signed-in user to check\n- screen_lock.password | Screen lock password requirement | Compliant | machine inactivity limit locks the session\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | DEVSEIS\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | LAPTOP-Y6NRCJU\\lmuller, LAPTOP-Y6NRCJU\\tnguyen\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (DEVSEIS\\asmith)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Non-Compliant | 3 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 6 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 76 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | AUOptions=2\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 104 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26100\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Non-Compliant | type NoSync, service Stopped/Manual\n- gdpr.backup_configured | Backup configured | Compliant | OneDrive folder backup (policy)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | chromeremotedesktop\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=1 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Compliant | Recall disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.complexity: Medium\n- lockout.duration: Low\n- firewall.domain.enabled: High\n- firewall.domain.inbound: High\n- firewall.domain.outbound: Low\n- firewall.private.enabled: High\n- firewall.private.inbound: High\n- firewall.private.outbound: Low\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.time_sync: Low\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.human_oversight: Medium\nSCORE (computed): compliant=26, assessed=53, percent=49\nTOP GAPS (computed, highest risk first):\n- firewall.domain.enabled | Domain firewall | High | off\n- firewall.domain.inbound | Domain inbound | High | allow\n- firewall.private.enabled | Private firewall | High | off\n- firewall.private.inbound | Private inbound | High | allow\n- firewall.public.inbound | Public inbound | High | allow\nFIX GUIDANCE (highest-risk gaps):\n- firewall.domain.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- firewall.private.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-Y6NRCJU - Microsoft Windows 11 Pro 10.0.26100\", \"score\": {\"compliant\": 26, \"assessed\": 53, \"percent\": 49}, \"counts\": {\"Compliant\": 26, \"Non-Compliant\": 27, \"NotApplicable\": 8, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.domain.enabled\", \"name\": \"Domain firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.private.enabled\", \"name\": \"Private firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.public.inbound\", \"name\": \"Public inbound\", \"risk\": \"High\", \"found\": \"allow\"}], \"summary\": \"LAPTOP-Y6NRCJU meets 26 of 53 assessed checks (49%). The most serious gaps are Domain firewall, Domain inbound and Private firewall. 3 items are pending.\", \"next_steps\": [\"Domain firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Private firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"Found: whoopsie error reporting on. Required: usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain outbound: does not meet the baseline\", \"finding\": \"Domain outbound is block instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 227 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-04 Cumulative Update for Windows 11 (2026-02-20)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Found: 227 days ago. Required: <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Gemini, Otter\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: Gemini, Otter. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 7 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 482 source packages checked against ubuntu:22.04 advisories. git 1:2.34.1-1ubuntu1.6 (1 CVE, high, actively exploited; fixed in 1:2.34.1-1ubuntu1.13; e.g. CVE-2025-48384); libwebp 1.2.2-2ubuntu0.22.04.1 (1 CVE, medium, actively exploited; fixed in 1.2.2-2ubuntu0.22.04.2; e.g. CVE-2023-4863); nvidia-graphics-drivers-510-server 510.73.05-0ubuntu0.22.04.1 (3 CVEs, high; fixed in 510.85.02-0ubuntu0.22.04.1; e.g. CVE-2022-31607, CVE-2022-31608); nvidia-graphics-drivers-390 390.151-0ubuntu0.22.04.1 (3 CVEs, high; fixed in 390.154-0ubuntu0.22.04.1; e.g. CVE-2022-31607, CVE-2022-31608); golang-1.18 1.18.1-1ubuntu1.2 (1 CVE, high; fixed in 1.18.1-1ubuntu1.3; e.g. CVE-2026-39821); and 1 more. 399 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"The device reports 7 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nHOSTNAME: dev-ubuntu-61\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=31, Non-Compliant=20, NotApplicable=4, Pending=1, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 8\n- password.history | Password history | Non-Compliant | 0\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=3, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Non-Compliant | not set\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Compliant | 3 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (sali)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 5 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 17 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 24.04.3 LTS\n- gdpr.security_logging | Security event logging | Non-Compliant | auditd not running, journal volatile\n- gdpr.time_sync | Time synchronisation | Non-Compliant | NTP service no, synchronised no\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.antivirus: High\n- malware.service: High\n- malware.signature_updates: Medium\n- malware.on_access: Medium\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- gdpr.security_logging: Medium\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.telemetry: Low\n- org.ropa: High\n- org.breach_procedure: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=31, assessed=51, percent=61\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.service | ClamAV malware protection service | High | not running\n- vuln.pending_updates | Pending security updates | High | 5 pending\n- gdpr.backup_configured | Backup configured | High | none\n- org.ropa | Record of processing activities | High | partial\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- vuln.pending_updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed."}, {"role": "assistant", "content": "{\"device\": \"dev-ubuntu-61 - Ubuntu 24.04.3 LTS 24.04\", \"score\": {\"compliant\": 31, \"assessed\": 51, \"percent\": 61}, \"counts\": {\"Compliant\": 31, \"Non-Compliant\": 20, \"NotApplicable\": 4, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"5 pending\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"partial\"}], \"summary\": \"dev-ubuntu-61 meets 31 of 51 assessed checks (61%). The most serious gaps are Antivirus, ClamAV malware protection service and Pending security updates. 1 items are pending.\", \"next_steps\": [\"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Pending security updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Public inbound: does not meet the baseline\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Jan\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Jan.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"Lockout observation period is no lockout configured instead of >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutwindow:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Removable media encryption is USB storage writable without enforced encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 168 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-04-16  02:44:01\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Found: 168 days ago. Required: <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Removable media encryption is external drives writable without enforced encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Public firewall: does not meet the baseline\", \"finding\": \"Public firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history: meets the baseline\", \"finding\": \"Found: 24. This satisfies the baseline requirement of >= 24.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"The device reports not set, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"The user reports assistant screen access on, but the baseline requires screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: auditd running, journal volatile\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging matches the required setting\", \"finding\": \"Security event logging is set to auditd running, journal volatile, in line with the required auditd running.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no passcode\nREQUIRED: passcode set\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Passcode: does not meet the baseline\", \"finding\": \"The user reports no passcode, but the baseline requires passcode set. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall\", \"finding\": \"Firewall is no active firewall instead of active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: download only\nREQUIRED: download and install on\nDETAIL: Updates wait for the user to install them.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic iOS updates: does not meet the baseline\", \"finding\": \"Automatic iOS updates is download only instead of download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Compliant\", \"title\": \"Human oversight for high-risk AI matches the required setting\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: DeepSeek, Meta AI\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Unapproved AI tools (shadow AI) is DeepSeek, Meta AI instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 6-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Compliant\", \"title\": \"Screen lock strength is compliant\", \"finding\": \"Screen lock strength is set to 6-digit PIN, in line with the required >= 6 digits or alphanumeric. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: restic-backup.timer.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Confirm the last backup date in: restic-backup.timer. Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Compliant\", \"title\": \"Firewall inbound policy matches the required setting\", \"finding\": \"The device reports deny, which meets the requirement (deny).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Minimum password length is 6 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iPadOS 26.5 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iPadOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates: does not meet the baseline\", \"finding\": \"The device reports iPadOS 26.5 (update to 26.7.1 available), but the baseline requires latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: unattended-upgrades installed=1, enabled=1\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates: meets the baseline\", \"finding\": \"Automatic security updates is set to unattended-upgrades installed=1, enabled=1, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 13.7.8 (macos)\nHOSTNAME: OPS-MacBook-20\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=19, Non-Compliant=27, NotApplicable=5, Pending=5, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | not set\n- malware.antivirus | Antivirus | Compliant | XProtect 5244\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5237, 30 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 600 idle, password delay 0\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after 900 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | mkhan\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (sali)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no recovery key\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 183 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 42 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | macOS 13.7.8\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Non-Compliant | Network Time: Off\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Non-Compliant | guest login 1, guest file sharing 0\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- lockout.duration: Low\n- lockout.observation: Low\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.os_supported: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.guest_account: Medium\n- gdpr.telemetry: Low\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=19, assessed=46, percent=41\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | mkhan\n- user.account_type | Signed-in user account type | High | Administrator (sali)\n- vuln.pending_updates | Pending security updates | High | 1 pending\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin."}, {"role": "assistant", "content": "{\"device\": \"OPS-MacBook-20 - macOS 13.7.8\", \"score\": {\"compliant\": 19, \"assessed\": 46, \"percent\": 41}, \"counts\": {\"Compliant\": 19, \"Non-Compliant\": 27, \"NotApplicable\": 5, \"Pending\": 5, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"mkhan\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (sali)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"1 pending\"}], \"summary\": \"OPS-MacBook-20 meets 19 of 46 assessed checks (41%). The most serious gaps are Firewall, Firewall inbound policy and Unauthorized employee administrator accounts. 5 items are pending.\", \"next_steps\": [\"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Answer the 5 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] How to install on a Mac: Open the downloaded .dmg file and drag Devseis Endpoint Auditor to Applications. The first time, macOS says it could not verify the app because it is not yet notarised: click Done, then open System Settings, Privacy & Security, and click Open Anyway. After that it opens normally. It needs macOS 13 or newer.\n[2] How to run an audit of a computer: A web page cannot read a computer's settings, so computers are audited with the free Devseis desktop app for Mac, Windows (beta) and Linux. Download it from the Download page, install it, click Run audit and enter the computer password once. The first run downloads the AI auditor; the audit then takes about 5 to 15 minutes and ends with a PDF report.\n[3] What is in the audit report: The report has a compliance score, the most important gaps, and for each problem: what was found, why it matters, the risk level, step-by-step fixes for that operating system, how to verify the fix, and the related ISO 27001, GDPR and EU AI Act references. It also lists AI tools found and their EU AI Act risk class.\n[4] Does the auditor need internet?: Only the first time, to download the AI auditor and the vulnerability list. After that audits and chat work offline. Pending security updates can only be checked online because they ask the vendor.\n\nMY AUDIT: none\n\nQUESTION: What should I fix first?"}, {"role": "assistant", "content": "I can't see your device from here, so I can't tell you yet. Run an audit with the free Devseis desktop app: it reads the settings, changes nothing, and takes about 5 to 15 minutes [2]. Then ask me again and I will answer from your real results."}], "meta": {"task": "chat", "os": "any", "status": "noaudit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Found: partial. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is compliant\", \"finding\": \"Unapproved AI tools (shadow AI) is set to none, in line with the required only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The user reports ChatGPT, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Malware security updates on macOS: Malware security updates on macOS. Why it matters: Signature updates must happen automatically so protection stays current without user action. Risk if missing: Medium - Manual updates are forgotten, leaving protection outdated. Requirement: Automatic signature/security-data updates are enabled. How to fix on macOS: (1) Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile. How to verify: Re-run the audit; signature updates should be automatic. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Security updates installed recently on macOS: Security updates installed recently on macOS. Why it matters: Recent installs show that patching is actually happening, not only configured. Risk if missing: High - A device not updated for a long time is likely missing critical security fixes. Requirement: Last successful update/package change within UPDATES_MAX_AGE_DAYS days. How to fix on macOS: (1) Install pending updates (System Settings > General > Software Update, or softwareupdate -ia). How to verify: Re-run the audit; the last update should be within 30 days. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience); GDPR Art. 32(1)(d) (Security of processing: regular testing and evaluation).\n[3] Outdated third-party applications on macOS: Outdated third-party applications on macOS. Why it matters: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Risk if missing: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages. Requirement: Installed third-party applications with a newer version available <= MAX_OUTDATED_APPS (Windows: winget; macOS: app versions; Linux: snap/flatpak and packages). Phones: automatic app updates are on. How to fix on macOS: (1) Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM. How to verify: Re-run the audit; outdated applications should be 0 or fewer. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Automatic security updates on macOS: Automatic security updates on macOS. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on macOS: (1) Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true). How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: macOS 14.7.1 (macos) on a Mac, Firefox 143; detected by the browser, security settings not checked.\n\nQUESTION: What steps do I take for automatic security updates on macOS?"}, {"role": "assistant", "content": "Security fixes must be installed automatically so that known vulnerabilities are closed quickly. To fix it on macOS [4]:\n1. Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\n\nTo check it worked: Re-run the audit; Automatic security updates should be Compliant."}], "meta": {"task": "chat", "os": "macos", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: screensharing\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is screensharing instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone is non-compliant\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: until an administrator unlocks\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Compliant\", \"title\": \"Account lockout duration matches the required setting\", \"finding\": \"Found: until an administrator unlocks. This satisfies the baseline requirement of >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates: does not meet the baseline\", \"finding\": \"The user reports manual, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after 900 seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"The device reports password after 900 seconds, but the baseline requires password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in bzserv.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Current state: date not readable. Confirm the last backup date in bzserv.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: restricted by profile\nREQUIRED: restricted by profile\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Compliant\", \"title\": \"Apple Intelligence external integrations restricted matches the required setting\", \"finding\": \"Apple Intelligence external integrations restricted is set to restricted by profile, in line with the required restricted by profile.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain outbound is non-compliant\", \"finding\": \"Found: allow. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Pending\", \"title\": \"Find My iPhone: needs a follow-up step\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answer ORG_DPIA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAI TOOLS FOUND: Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude, Cursor, Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"The device reports Claude, Cursor, Claude Code, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 12 (android)\nHOSTNAME: legal-galaxy-44\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=19, Non-Compliant=8, NotApplicable=0, Pending=3, Error=0\nEVIDENCE: 0 detected by the browser, 22 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Non-Compliant | Screen timeout 10 minutes\n- mobile.passcode_strength | Screen lock strength | Compliant | 6-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 12 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-10 (0 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 12\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | ChatGPT\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.timeout: Medium\n- vuln.outdated_apps: Medium\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- aiact.unapproved: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=19, assessed=27, percent=70\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- gdpr.os_supported | Supported operating system version | High | Android 12\n- gdpr.backup_configured | Google backup | High | no backup\n- screen_lock.timeout | Screen timeout | Medium | Screen timeout 10 minutes\n- vuln.outdated_apps | App updates | Medium | automatic app updates off\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- gdpr.os_supported: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\n- gdpr.backup_configured: Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps."}, {"role": "assistant", "content": "{\"device\": \"legal-galaxy-44 - Android 12\", \"score\": {\"compliant\": 19, \"assessed\": 27, \"percent\": 70}, \"counts\": {\"Compliant\": 19, \"Non-Compliant\": 8, \"NotApplicable\": 0, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 12\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Google backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Screen timeout\", \"risk\": \"Medium\", \"found\": \"Screen timeout 10 minutes\"}, {\"check_id\": \"vuln.outdated_apps\", \"name\": \"App updates\", \"risk\": \"Medium\", \"found\": \"automatic app updates off\"}], \"summary\": \"legal-galaxy-44 meets 19 of 27 assessed checks (70%). The most serious gaps are Google Play Protect, Supported operating system version and Google backup. 3 items are pending. 22 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Supported operating system version: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\", \"Google backup: Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off for: sali, asmith\nREQUIRED: on\nDETAIL: Users checked: sali, asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock is non-compliant\", \"finding\": \"The device reports off for: sali, asmith, but the baseline requires on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.5 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iOS security updates\", \"finding\": \"iOS security updates is iOS 26.5 (update to 26.7.1 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"USB storage device access is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off for: rschmit\nREQUIRED: on\nDETAIL: Users checked: rschmit\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Screen lock is off for: rschmit instead of on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nHOSTNAME: legal-lt-62\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=29, Non-Compliant=17, NotApplicable=2, Pending=7, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Compliant | minclass=3, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Non-Compliant | no signature database\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: lmuller\n- usb.device_access | USB storage device access | Error | not read\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | akaya, jdoe\n- user.account_type | Signed-in user account type | Compliant | Standard (lmuller)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 38 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 16 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 35 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 64 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Debian GNU/Linux 12 (bookworm)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | restic-backup.timer\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Jan\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Jan\n- aiact.risk_classification | AI system risk classification | Pending | Jan\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- lockout.duration: Low\n- malware.signature_updates: Medium\n- malware.signatures: High\n- malware.on_access: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- screen_lock.password: Medium\n- admin.unauthorized: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- aiact.unapproved: Medium\n- org.ai_register: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=29, assessed=46, percent=63\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Malware signatures | High | no signature database\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\n- admin.unauthorized | Unauthorized employee administrator accounts | High | akaya, jdoe\n- vuln.pending_updates | Pending security updates | High | 38 pending\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\n- firewall.enabled: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\n- firewall.inbound: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT)."}, {"role": "assistant", "content": "{\"device\": \"legal-lt-62 - Debian GNU/Linux 12 (bookworm) 12\", \"score\": {\"compliant\": 29, \"assessed\": 46, \"percent\": 63}, \"counts\": {\"Compliant\": 29, \"Non-Compliant\": 17, \"NotApplicable\": 2, \"Pending\": 7, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"no signature database\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"akaya, jdoe\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"38 pending\"}], \"summary\": \"legal-lt-62 meets 29 of 46 assessed checks (63%). The most serious gaps are Malware signatures, Firewall and Firewall inbound policy. 1 checks could not be read. 7 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Firewall: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\", \"Firewall inbound policy: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\", \"Answer the 4 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Character.AI, Meta AI, Otter\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Unapproved AI tools (shadow AI) is Character.AI, Meta AI, Otter instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Domain firewall: does not meet the baseline\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\", \"For a stand-alone device: net accounts /maxpwage:365\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Ollama\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: Ollama. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: unattended-upgrades installed=1, enabled=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates: does not meet the baseline\", \"finding\": \"The device reports unattended-upgrades installed=1, enabled=0, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: smb\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"Found: smb. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rooted or bootloader unlocked\nREQUIRED: not rooted\nDETAIL: Android security protections are weakened on a rooted phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Not rooted\", \"finding\": \"The user reports rooted or bootloader unlocked, but the baseline requires not rooted. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: smb\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"Remote access services restricted is smb instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Gemini\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is Gemini instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: macOS 26.0.1\nREQUIRED: macOS 14 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (macos):\n- Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version: meets the baseline\", \"finding\": \"Found: macOS 26.0.1. This satisfies the baseline requirement of macOS 14 or later.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: XProtect 5376, 61 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware protection signatures: does not meet the baseline\", \"finding\": \"The device reports XProtect 5376, 61 days old, but the baseline requires <= 30 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 317 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 17.7 checked against NVD iOS records. iOS 17.7 (317 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2024-27874, CVE-2024-40856). 547 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in iOS\", \"finding\": \"Found: 317 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted: does not meet the baseline\", \"finding\": \"The user reports assistant screen access on, but the baseline requires screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: recovery key present\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Error\", \"title\": \"Disk encryption recovery: could not be checked\", \"finding\": \"The collector could not read this setting (Requires root: run the collector with sudo.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, None\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker protection\", \"finding\": \"Found: protection Off, None. Required: protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] AI system risk classification on Windows: AI system risk classification on Windows. Why it matters: Each AI system must be checked against the banned practices and the high-risk uses in Annex III. Risk if missing: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards. Requirement: Each AI tool found is classified (prohibited / high-risk / limited / minimal). Done by the auditor model: Pending in the collector output, NotApplicable when no AI tool is found. How to fix on Windows: (1) Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose. How to verify: Every AI tool in the inventory has a classification in the AI register. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); EU AI Act Art. 5 (Prohibited AI practices); EU AI Act Art. 6 (Classification rules for high-risk AI systems); EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)).\n[2] Unapproved AI tools (shadow AI) on Windows: Unapproved AI tools (shadow AI) on Windows. Why it matters: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Risk if missing: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations. Requirement: Every AI tool found is listed in APPROVED_AI_TOOLS. How to fix on Windows: (1) Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it. (2) Publish an AI acceptable-use policy so staff know which tools are allowed. (3) Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control. How to verify: Re-run the audit; Unapproved AI tools should read none. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 28 (Processor); GDPR Art. 44 (General principle for transfers); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[3] Windows Recall / Copilot data analysis disabled on Windows: Windows Recall / Copilot data analysis disabled on Windows. Why it matters: Built-in OS AI features can capture screen content or send text to external AI services. Risk if missing: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose. Requirement: OS-level AI features that capture or send user data are disabled by policy. NotApplicable where the OS has none. Phones: external AI integrations (ChatGPT in Apple Intelligence, assistant access to screen content) are off unless approved. How to fix on Windows: (1) Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved. How to verify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[4] AI software inventory on Windows: AI software inventory on Windows. Why it matters: Deployers must know which AI systems they use before they can classify and govern them. Risk if missing: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools. Requirement: AI apps, local AI services and local model folders are listed (Compliant = inventory taken). How to fix on Windows: (1) Add the AI tools found on this device to the organisation's AI register, with owner and purpose. How to verify: The AI register lists every tool in the inventory. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); GDPR Art. 30 (Records of processing activities); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n\nMY AUDIT: none\n\nMY DEVICE: Windows 10 (windows) on a Windows PC (ARM), Firefox 143; detected by the browser, security settings not checked.\n\nQUESTION: My audit says 'AI software inventory' is non-compliant on my Windows. What should I do?"}, {"role": "assistant", "content": "Deployers must know which AI systems they use before they can classify and govern them. To fix it on Windows [4]:\n1. Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\n\nTo check it worked: The AI register lists every tool in the inventory."}], "meta": {"task": "chat", "os": "windows", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ewagner\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"Found: ewagner. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Screen timeout 10 minutes\nREQUIRED: <= 300 seconds\nDETAIL: The phone stays unlocked for 10 minutes when not in use.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen timeout is non-compliant\", \"finding\": \"Found: Screen timeout 10 minutes. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"Screen lock timeout is no machine inactivity limit; screen saver off instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"The user reports none, which meets the requirement (only: none). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"The device reports USB storage writable without enforced encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: external AI off unless approved\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Pending\", \"title\": \"Apple Intelligence external integrations restricted is pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 8\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"The device reports 8, but the baseline requires >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: disabled\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Error\", \"title\": \"Guest account disabled could not be read\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Grammarly, Microsoft Copilot\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: ChatGPT, Grammarly, Microsoft Copilot. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: ssh, teamviewer\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Found: ssh, teamviewer. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: none enabled\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Compliant\", \"title\": \"Remote access services restricted is compliant\", \"finding\": \"The device reports none enabled, which meets the requirement (only: none).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 105 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 18 applications inventoried, 12 recognised (23 common applications covered). Adobe Acrobat Reader 25.001.20529 (49 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-43550, CVE-2025-43573); Visual Studio Code 1.119.1 (23 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-40376, CVE-2026-45482); VLC 2.1.0 (22 CVEs, critical; fixed in 3.0.20; e.g. CVE-2014-9625, CVE-2014-9626); Safari 26.5.2 (7 CVEs, high; fixed in 26.6.1; e.g. CVE-2026-43794, CVE-2026-64713); Microsoft Teams 7.10.1 (4 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2025-49737, CVE-2025-53783). 71 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 105 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: check, download and install automatic\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates: meets the baseline\", \"finding\": \"Automatic security updates is set to check, download and install automatic, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no screen lock (Swipe/None)\nREQUIRED: PIN, password or pattern\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"Screen lock is no screen lock (Swipe/None) instead of PIN, password or pattern. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an enabled administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"Found: none of the approved accounts is an enabled administrator. Required: itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, None\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker protection\", \"finding\": \"Found: protection Off, None. Required: protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: >= 6 months for high-risk AI\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"NotApplicable\", \"title\": \"AI usage log retention does not apply here\", \"finding\": \"AI usage log retention is not assessed here: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: passcode set\nREQUIRED: passcode set\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Passcode: meets the baseline\", \"finding\": \"Found: passcode set. This satisfies the baseline requirement of passcode set. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 90; accounts: jdoe (99999)\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"Password expiration is PASS_MAX_DAYS 90; accounts: jdoe (99999) instead of 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Human oversight for high-risk AI is no instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_pwquality not enabled\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Password complexity is pam_pwquality not enabled instead of 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Found: no machine inactivity limit; screen saver off. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: letters and numbers required\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Compliant\", \"title\": \"Password complexity is compliant\", \"finding\": \"Found: letters and numbers required. This satisfies the baseline requirement of enforced.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age is non-compliant\", \"finding\": \"Found: 0 days. Required: >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.5.2 (macos)\nHOSTNAME: DEV-MacBook-28\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=46, Non-Compliant=4, NotApplicable=5, Pending=1, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5290\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5322, 20 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 180 idle, password delay 300\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 0 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (mkhan)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 16 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.5.2\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Non-Compliant | Network Time: Off\n- gdpr.backup_configured | Backup configured | Compliant | bzserv\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | filesharing\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | restricted by profile\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- screen_lock.timeout: Medium\n- gdpr.time_sync: Low\n- gdpr.remote_access: High\n- org.human_oversight: Medium\nSCORE (computed): compliant=46, assessed=50, percent=92\nTOP GAPS (computed, highest risk first):\n- gdpr.remote_access | Remote access services restricted | High | filesharing\n- screen_lock.timeout | Screen lock timeout | Medium | 180 idle, password delay 300\n- org.human_oversight | Human oversight for high-risk AI | Medium | partial\n- gdpr.time_sync | Time synchronisation | Low | Network Time: Off\nFIX GUIDANCE (highest-risk gaps):\n- gdpr.remote_access: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\n- screen_lock.timeout: Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\n- org.human_oversight: For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used."}, {"role": "assistant", "content": "{\"device\": \"DEV-MacBook-28 - macOS 26.5.2\", \"score\": {\"compliant\": 46, \"assessed\": 50, \"percent\": 92}, \"counts\": {\"Compliant\": 46, \"Non-Compliant\": 4, \"NotApplicable\": 5, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"filesharing\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Screen lock timeout\", \"risk\": \"Medium\", \"found\": \"180 idle, password delay 300\"}, {\"check_id\": \"org.human_oversight\", \"name\": \"Human oversight for high-risk AI\", \"risk\": \"Medium\", \"found\": \"partial\"}, {\"check_id\": \"gdpr.time_sync\", \"name\": \"Time synchronisation\", \"risk\": \"Low\", \"found\": \"Network Time: Off\"}], \"summary\": \"DEV-MacBook-28 meets 46 of 50 assessed checks (92%). The most serious gaps are Remote access services restricted. 1 items are pending.\", \"next_steps\": [\"Remote access services restricted: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\", \"Screen lock timeout: Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\", \"Human oversight for high-risk AI: For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iOS 27.0\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Safari 27.0).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version: meets the baseline\", \"finding\": \"Supported operating system version is set to iOS 27.0, in line with the required >= iOS 18.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 12 (android)\nHOSTNAME: mkt-galaxy-08\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=17, Non-Compliant=3, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 2 detected by the browser, 20 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 30 seconds\n- mobile.passcode_strength | Screen lock strength | Compliant | 6-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 12 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-09 (1 month old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 12\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | phone maker cloud backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | Microsoft Copilot\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Microsoft Copilot\n- aiact.risk_classification | AI system risk classification | Pending | Microsoft Copilot\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- gdpr.os_supported: High\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=17, assessed=20, percent=85\nTOP GAPS (computed, highest risk first):\n- gdpr.os_supported | Supported operating system version | High | Android 12\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | Microsoft Copilot\n- aiact.os_ai_features | AI assistant access restricted | Medium | assistant screen access on\nFIX GUIDANCE (highest-risk gaps):\n- gdpr.os_supported: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- aiact.os_ai_features: In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it."}, {"role": "assistant", "content": "{\"device\": \"mkt-galaxy-08 - Android 12\", \"score\": {\"compliant\": 17, \"assessed\": 20, \"percent\": 85}, \"counts\": {\"Compliant\": 17, \"Non-Compliant\": 3, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 12\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"Microsoft Copilot\"}, {\"check_id\": \"aiact.os_ai_features\", \"name\": \"AI assistant access restricted\", \"risk\": \"Medium\", \"found\": \"assistant screen access on\"}], \"summary\": \"mkt-galaxy-08 meets 17 of 20 assessed checks (85%). The most serious gaps are Supported operating system version. 10 items are pending. 20 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Supported operating system version: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"AI assistant access restricted: In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LAPTOP-0OAT7LM\\akaya\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts is non-compliant\", \"finding\": \"The device reports LAPTOP-0OAT7LM\\\\akaya, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: chromeremotedesktop, smb\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"Found: chromeremotedesktop, smb. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store: does not meet the baseline\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store is non-compliant\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked: does not meet the baseline\", \"finding\": \"Notification previews hidden when locked is Always instead of When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: auditd not running, journal volatile\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security event logging\", \"finding\": \"Security event logging is auditd not running, journal volatile instead of auditd running. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: LEGAL-MacBook-12\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=31, Non-Compliant=13, NotApplicable=3, Pending=7, Error=2\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5233\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5364, 7 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall on, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 60 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | rschmit\n- user.account_type | Signed-in user account type | Compliant | Standard (pmartin)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Error | not read\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 69 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 269 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 82 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | Time Machine\n- gdpr.backup_recent | Recent backup | Non-Compliant | 83 days ago\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Error | not read\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude\n- aiact.risk_classification | AI system risk classification | Pending | Claude\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- firewall.inbound: High\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=31, assessed=44, percent=70\nTOP GAPS (computed, highest risk first):\n- firewall.inbound | Firewall inbound policy | High | firewall on, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | rschmit\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 269 critical/high\n- gdpr.updates_recent | Security updates installed recently | High | 82 days ago\n- usb.write | USB storage write | Medium | allow\nFIX GUIDANCE (highest-risk gaps):\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"LEGAL-MacBook-12 - macOS 26.0.1\", \"score\": {\"compliant\": 31, \"assessed\": 44, \"percent\": 70}, \"counts\": {\"Compliant\": 31, \"Non-Compliant\": 13, \"NotApplicable\": 3, \"Pending\": 7, \"Error\": 2}, \"top_gaps\": [{\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall on, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"rschmit\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"269 critical/high\"}, {\"check_id\": \"gdpr.updates_recent\", \"name\": \"Security updates installed recently\", \"risk\": \"High\", \"found\": \"82 days ago\"}, {\"check_id\": \"usb.write\", \"name\": \"USB storage write\", \"risk\": \"Medium\", \"found\": \"allow\"}], \"summary\": \"LEGAL-MacBook-12 meets 31 of 44 assessed checks (70%). The most serious gaps are Firewall inbound policy, Unauthorized employee administrator accounts and Known vulnerabilities in installed software. 2 checks could not be read. 7 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 5 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: Defender in Passive Mode\nREQUIRED: on\nDETAIL: Another antivirus is primary: Sophos Endpoint\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"NotApplicable\", \"title\": \"Tamper protection: not assessed on this device\", \"finding\": \"Not applicable on this device: Another antivirus is primary: Sophos Endpoint.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: Recall not available on this device\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Windows Recall / Copilot data analysis disabled: not assessed on this device\", \"finding\": \"This check does not apply: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: disabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled is compliant\", \"finding\": \"Found: disabled. This satisfies the baseline requirement of disabled.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Alpaca (Ollama client)\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: Alpaca (Ollama client). Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (windows):\n- Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read is compliant\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 14\nREQUIRED: >= 14\nDETAIL: pwpolicy global account policy\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (macos):\n- Deploy a Passcode configuration profile through your MDM with minLength = 14.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Compliant\", \"title\": \"Minimum password length: meets the baseline\", \"finding\": \"Minimum password length is set to 14, in line with the required >= 14.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answer ORG_AI_TRANSPARENCY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2025-08 (14 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2025-08 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Android security update\", \"finding\": \"Android security update is security patch 2025-08 (14 months old) instead of <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Block or allow external storage with your MDM (Restrictions / device control) or an endpoint DLP product.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 18.7 (ios)\nHOSTNAME: MKT-iPhone-26\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=21, Non-Compliant=4, NotApplicable=0, Pending=4, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 1 minute\n- mobile.passcode_strength | Passcode strength | Compliant | 7+ digit numeric code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store and employer-managed apps\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 18.7 (update to 18.7.10 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 207 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 18.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | ChatGPT\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Pending | not answered\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- aiact.unapproved: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=21, assessed=25, percent=84\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | iOS security updates | High | iOS 18.7 (update to 18.7.10 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 207 critical/high\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | ChatGPT\n- org.human_oversight | Human oversight for high-risk AI | Medium | partial\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it."}, {"role": "assistant", "content": "{\"device\": \"MKT-iPhone-26 - iOS 18.7\", \"score\": {\"compliant\": 21, \"assessed\": 25, \"percent\": 84}, \"counts\": {\"Compliant\": 21, \"Non-Compliant\": 4, \"NotApplicable\": 0, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 18.7 (update to 18.7.10 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"207 critical/high\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"ChatGPT\"}, {\"check_id\": \"org.human_oversight\", \"name\": \"Human oversight for high-risk AI\", \"risk\": \"Medium\", \"found\": \"partial\"}], \"summary\": \"MKT-iPhone-26 meets 21 of 25 assessed checks (84%). The most serious gaps are iOS security updates and Known vulnerabilities in iOS. 4 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 353 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 737 source packages checked against debian:12 advisories. activemq 5.14.3-3 (5 CVEs, critical, actively exploited; fixed in 5.17.2+dfsg-2+deb12u1; e.g. CVE-2018-11775, CVE-2019-0222); wireshark 2.2.7-1 (159 CVEs, critical; fixed in 4.0.17-0+deb12u3; e.g. CVE-2017-11406, CVE-2017-11407); tiff 4.0.3-1 (77 CVEs, critical; fixed in 4.5.0-6+deb12u4; e.g. CVE-2014-8129, CVE-2015-7554); wordpress 2.8.3-1 (50 CVEs, critical; fixed in 6.1.6+dfsg1-0+deb12u1; e.g. CVE-2016-2221, CVE-2016-2222); mosquitto 1.5.4-1 (10 CVEs, critical; fixed in 2.0.11-1.2+deb12u2; e.g. CVE-2018-12550, CVE-2018-12551); and 17 more. 429 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"Found: 353 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: dnf-automatic timer on, apply_updates=yes\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates matches the required setting\", \"finding\": \"Automatic security updates is set to dnf-automatic timer on, apply_updates=yes, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store: does not meet the baseline\", \"finding\": \"The user reports apps from outside the App Store, but the baseline requires App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no recovery key\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption recovery\", \"finding\": \"Disk encryption recovery is no recovery key instead of recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Compliant\", \"title\": \"AI register matches the required setting\", \"finding\": \"AI register is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 42 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-08-25 01:49:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"The device reports 42 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"Breach response procedure is partial instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history matches the required setting\", \"finding\": \"Password history is set to 24, in line with the required >= 24.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: filesharing\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"Found: filesharing. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: auditd not running, journal persistent\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging is non-compliant\", \"finding\": \"Security event logging is auditd not running, journal persistent instead of auditd running. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no FileVault\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"Found: no FileVault. Required: recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted: does not meet the baseline\", \"finding\": \"Apple Intelligence external integrations restricted is not restricted instead of restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Security updates installed recently on Linux: Security updates installed recently on Linux. Why it matters: Recent installs show that patching is actually happening, not only configured. Risk if missing: High - A device not updated for a long time is likely missing critical security fixes. Requirement: Last successful update/package change within UPDATES_MAX_AGE_DAYS days. How to fix on Linux: (1) Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors. How to verify: Re-run the audit; the last update should be within 30 days. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience); GDPR Art. 32(1)(d) (Security of processing: regular testing and evaluation).\n[2] Automatic security updates on Linux: Automatic security updates on Linux. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on Linux: (1) Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer. How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Malware signature updates on Linux: Malware signature updates on Linux. Why it matters: Signature updates must happen automatically so protection stays current without user action. Risk if missing: Medium - Manual updates are forgotten, leaving protection outdated. Requirement: Automatic signature/security-data updates are enabled. How to fix on Linux: (1) systemctl enable --now clamav-freshclam. How to verify: Re-run the audit; signature updates should be automatic. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Outdated third-party applications on Linux: Outdated third-party applications on Linux. Why it matters: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Risk if missing: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages. Requirement: Installed third-party applications with a newer version available <= MAX_OUTDATED_APPS (Windows: winget; macOS: app versions; Linux: snap/flatpak and packages). Phones: automatic app updates are on. How to fix on Linux: (1) Update packages and app stores: apt upgrade, snap refresh, flatpak update. How to verify: Re-run the audit; outdated applications should be 0 or fewer. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer (x86-64), Chrome 141; detected by the browser, security settings not checked.\n\nQUESTION: Why does automatic security updates matter?"}, {"role": "assistant", "content": "Security fixes must be installed automatically so that known vulnerabilities are closed quickly. If it is missing, the risk is high: Unpatched devices are the most common way attackers get in. [2]"}], "meta": {"task": "chat", "os": "linux", "status": "why", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 175 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-04-13  05:50:39\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Security updates installed recently is 175 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup: does not meet the baseline\", \"finding\": \"The user reports no backup, but the baseline requires backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain outbound: does not meet the baseline\", \"finding\": \"Found: allow. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Character.AI, Meta AI, Otter\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Character.AI, Meta AI, Otter. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: <= 0 pending\nDETAIL: No answer within 300 seconds.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Error\", \"title\": \"Pending security updates: could not be checked\", \"finding\": \"No evidence was collected for this check (No answer within 300 seconds.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 12 (android)\nHOSTNAME: hr-motorola-58\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=20, Non-Compliant=5, NotApplicable=2, Pending=3, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 30 seconds\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | 4-digit PIN\n- mobile.jailbreak_root | Not rooted | Non-Compliant | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 12 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-09 (1 month old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 12\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | no notifications on lock screen\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Pending | not answered\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- gdpr.os_supported: High\n- org.breach_procedure: High\n- org.ai_transparency: Low\nSCORE (computed): compliant=20, assessed=25, percent=80\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not rooted | High | rooted or bootloader unlocked\n- gdpr.os_supported | Supported operating system version | High | Android 12\n- org.breach_procedure | Breach response procedure | High | partial\n- mobile.passcode_strength | Screen lock strength | Medium | 4-digit PIN\n- org.ai_transparency | AI transparency to users | Low | partial\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\n- gdpr.os_supported: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\n- org.breach_procedure: Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise."}, {"role": "assistant", "content": "{\"device\": \"hr-motorola-58 - Android 12\", \"score\": {\"compliant\": 20, \"assessed\": 25, \"percent\": 80}, \"counts\": {\"Compliant\": 20, \"Non-Compliant\": 5, \"NotApplicable\": 2, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not rooted\", \"risk\": \"High\", \"found\": \"rooted or bootloader unlocked\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 12\"}, {\"check_id\": \"org.breach_procedure\", \"name\": \"Breach response procedure\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Screen lock strength\", \"risk\": \"Medium\", \"found\": \"4-digit PIN\"}, {\"check_id\": \"org.ai_transparency\", \"name\": \"AI transparency to users\", \"risk\": \"Low\", \"found\": \"partial\"}], \"summary\": \"hr-motorola-58 meets 20 of 25 assessed checks (80%). The most serious gaps are Not rooted, Supported operating system version and Breach response procedure. 3 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not rooted: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\", \"Supported operating system version: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\", \"Breach response procedure: Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: unified logging active\nREQUIRED: security events logged\nDETAIL: macOS unified log records authentication events; retention is managed by the OS.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (macos):\n- Unified logging is built in; forward security logs to your log platform if long retention is required.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging matches the required setting\", \"finding\": \"The device reports unified logging active, which meets the requirement (security events logged).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: minclass=0, required classes=4\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Compliant\", \"title\": \"Password complexity is compliant\", \"finding\": \"Password complexity is set to minclass=0, required classes=4, in line with the required 3 or more character classes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: LUKS2 aes-xts-plain64 512-bit key\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"Disk encryption protection matches the required setting\", \"finding\": \"The device reports LUKS2 aes-xts-plain64 512-bit key, which meets the requirement (active, strong cipher).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 156 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 24 applications inventoried, 13 recognised (23 common applications covered). Adobe Acrobat Reader 25.001.20997 (40 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2026-27220, CVE-2026-27278); Microsoft Teams 1.6.00.26463 (7 CVEs, high, actively exploited; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330); Firefox 153.3.0 (57 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100756, CVE-2026-100767); Visual Studio Code 1.119.1 (23 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-40376, CVE-2026-45482); VLC 2.1.6 (16 CVEs, critical; fixed in 3.0.20; e.g. CVE-2016-5108, CVE-2017-17670); and 3 more. 52 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 156 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: ClamAV malware protection service\", \"finding\": \"ClamAV malware protection service is not running instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Compliant\", \"title\": \"AI transparency to users: meets the baseline\", \"finding\": \"AI transparency to users is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 50\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Found: allow. Required: deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 152 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 24 applications inventoried, 14 recognised (31 common applications covered). WinRAR 7.00 (64-bit) 7.00 (2 CVEs, high, actively exploited; fixed in 7.13; e.g. CVE-2025-6218, CVE-2025-8088); Mozilla Firefox (x64 en-US) 153.2.0 (111 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100756, CVE-2026-100767); Microsoft Visual Studio Code (User) 1.119.1 (23 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-40376, CVE-2026-45482); Microsoft Edge 151.0.4129.107 (6 CVEs, high; fixed in 153.0.4234.46; e.g. CVE-2026-69486, CVE-2026-70341); Microsoft Teams 1.6.00.26474 (6 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330); and 3 more. 42 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 152 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Microsoft Copilot\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Current state: Microsoft Copilot. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 0 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-01 Cumulative Update for Windows 11 (2026-10-03)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Compliant\", \"title\": \"Security updates installed recently: meets the baseline\", \"finding\": \"The device reports 0 days ago, which meets the requirement (<= 30 days ago).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 207 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 18.7 checked against NVD iOS records. iOS 18.7 (207 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2025-14174, CVE-2025-31255). 356 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS: does not meet the baseline\", \"finding\": \"Known vulnerabilities in iOS is 207 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: File History (jdoe)\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Compliant\", \"title\": \"Backup configured is compliant\", \"finding\": \"The device reports File History (jdoe), which meets the requirement (backup configured).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 13 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 16 applications inventoried, 12 recognised (31 common applications covered). VLC media player 2.2.6 (13 CVEs, critical; fixed in 3.0.20; e.g. CVE-2017-10699, CVE-2017-17670). 3 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"The device reports 13 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery: does not meet the baseline\", \"finding\": \"BitLocker recovery is key protectors: Tpm instead of recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; no signed-in user to check\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock is non-compliant\", \"finding\": \"Found: no machine inactivity limit; no signed-in user to check. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: lmuller\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"Unauthorized employee administrator accounts is lmuller instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"The device reports not set, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public inbound\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 15.7.2 (macos)\nHOSTNAME: FIN-MacBook-10\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=36, Non-Compliant=11, NotApplicable=3, Pending=5, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout\n- malware.antivirus | Antivirus | Compliant | XProtect 5328\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5372, 16 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 60 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Error | not read\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | ewagner\n- user.account_type | Signed-in user account type | Compliant | Standard (rschmit)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 45 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 220 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 4 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 15.7.2\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | ArqAgent\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | screensharing\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Jan, Hugging Face models, Cursor\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Jan, Hugging Face models, Cursor\n- aiact.risk_classification | AI system risk classification | Pending | Jan, Hugging Face models, Cursor\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | restricted by profile\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- lockout.observation: Low\n- usb.write: Medium\n- admin.unauthorized: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.processor_agreements: High\nSCORE (computed): compliant=36, assessed=47, percent=77\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee administrator accounts | High | ewagner\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 220 critical/high\n- gdpr.remote_access | Remote access services restricted | High | screensharing\n- org.ropa | Record of processing activities | High | no\n- org.processor_agreements | Data processing agreements | High | no\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.remote_access: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host)."}, {"role": "assistant", "content": "{\"device\": \"FIN-MacBook-10 - macOS 15.7.2\", \"score\": {\"compliant\": 36, \"assessed\": 47, \"percent\": 77}, \"counts\": {\"Compliant\": 36, \"Non-Compliant\": 11, \"NotApplicable\": 3, \"Pending\": 5, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"ewagner\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"220 critical/high\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"screensharing\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"no\"}, {\"check_id\": \"org.processor_agreements\", \"name\": \"Data processing agreements\", \"risk\": \"High\", \"found\": \"no\"}], \"summary\": \"FIN-MacBook-10 meets 36 of 47 assessed checks (77%). The most serious gaps are Unauthorized employee administrator accounts, Known vulnerabilities in installed software and Remote access services restricted. 1 checks could not be read. 5 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Remote access services restricted: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: set automatically\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Pending\", \"title\": \"Date & time set automatically is pending\", \"finding\": \"The user could not find or confirm this setting. Current state: not answered. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Compliant\", \"title\": \"Firewall outbound policy: meets the baseline\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock is non-compliant\", \"finding\": \"Found: Auto-Lock Never. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nHOSTNAME: hr-dev-49\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=24, Non-Compliant=25, NotApplicable=2, Pending=5, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 10\n- password.history | Password history | Non-Compliant | 0\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 90; accounts: jdoe (99999)\n- password.min_age | Minimum password age | Non-Compliant | PASS_MIN_DAYS 0\n- password.complexity | Password complexity | Non-Compliant | minclass=0, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Compliant | 3 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Non-Compliant | off for: jdoe\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: jdoe\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | jdoe\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (jdoe)\n- encryption.enabled | Encryption | Non-Compliant | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | not encrypted\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 2 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 26.04.1 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Non-Compliant | NTP service no, synchronised no\n- gdpr.backup_configured | Backup configured | Compliant | Timeshift\n- gdpr.backup_recent | Recent backup | Non-Compliant | 70 days ago\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | smb\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | GPT4All\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | GPT4All\n- aiact.risk_classification | AI system risk classification | Pending | GPT4All\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.expiration: Low\n- password.min_age: Low\n- password.complexity: Medium\n- malware.signature_updates: Medium\n- malware.on_access: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- gdpr.time_sync: Low\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.processor_agreements: High\nSCORE (computed): compliant=24, assessed=49, percent=49\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\n- admin.unauthorized | Unauthorized employee administrator accounts | High | jdoe\n- user.account_type | Signed-in user account type | High | Administrator (jdoe)\n- encryption.enabled | Encryption | High | root filesystem not encrypted\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\n- firewall.inbound: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d."}, {"role": "assistant", "content": "{\"device\": \"hr-dev-49 - Ubuntu 26.04.1 LTS 26.04\", \"score\": {\"compliant\": 24, \"assessed\": 49, \"percent\": 49}, \"counts\": {\"Compliant\": 24, \"Non-Compliant\": 25, \"NotApplicable\": 2, \"Pending\": 5, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"jdoe\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (jdoe)\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"root filesystem not encrypted\"}], \"summary\": \"hr-dev-49 meets 24 of 49 assessed checks (49%). The most serious gaps are Firewall, Firewall inbound policy and Unauthorized employee administrator accounts. 5 items are pending.\", \"next_steps\": [\"Firewall: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\", \"Firewall inbound policy: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Answer the 3 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store: does not meet the baseline\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: The macOS application firewall filters inbound connections only.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (macos):\n- The macOS application firewall does not filter outbound traffic; use a network filter product if outbound blocking is required.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Compliant\", \"title\": \"Firewall outbound policy matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iPadOS 26.5 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iPadOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates: does not meet the baseline\", \"finding\": \"Found: iPadOS 26.5 (update to 26.7.1 available). Required: latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no passcode\nREQUIRED: passcode set\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Passcode: does not meet the baseline\", \"finding\": \"The user reports no passcode, but the baseline requires passcode set. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Compliant\", \"title\": \"Unauthorized employee administrator accounts matches the required setting\", \"finding\": \"The device reports none, which meets the requirement (none).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: download only\nREQUIRED: download and install on\nDETAIL: Updates wait for the user to install them.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic iOS updates is non-compliant\", \"finding\": \"Automatic iOS updates is download only instead of download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates is non-compliant\", \"finding\": \"Found: freshclam not running. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Find My Device\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Pending\", \"title\": \"Find My Device is pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in bzserv.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Confirm the last backup date in bzserv. Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no passcode\nREQUIRED: passcode set\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Passcode: does not meet the baseline\", \"finding\": \"The user reports no passcode, but the baseline requires passcode set. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Ollama\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: Ollama. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI register\", \"finding\": \"Found: no. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: On-access malware protection\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Compliant\", \"title\": \"Find My iPhone is compliant\", \"finding\": \"The user reports on, which meets the requirement (on). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 20\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"Account lockout threshold is 20 instead of 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Notification previews hidden when locked\", \"finding\": \"Notification previews hidden when locked is Always instead of When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: download and install on\nDETAIL: Security updates are not installed automatically.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic iOS updates: does not meet the baseline\", \"finding\": \"Found: off. Required: download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: off\nREQUIRED: off\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Compliant\", \"title\": \"USB debugging off matches the required setting\", \"finding\": \"The user reports off, which meets the requirement (off). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"The device reports allow (no firewall), but the baseline requires deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is anydesk instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rschmit\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Unauthorized employee administrator accounts is rschmit instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 353 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 528 packages checked against rhel:8 advisories. golang-docs 0:1.19.6-1.module+el8.8.0+18289+edd6c8b6 (75 CVEs, critical, actively exploited; fixed in 0:1.26.7-1.module+el8.10.0+24751+b3673b3e; e.g. CVE-2023-24532, CVE-2023-24534); kernel-abi-whitelists 0:4.18.0-80.1.2.el8_0 (182 CVEs, high, actively exploited; fixed in 0:4.18.0-240.22.1.el8_3; e.g. CVE-2015-1593, CVE-2018-16884); tomcat-admin-webapps 1:9.0.87-1.el8_10.3 (42 CVEs, high, actively exploited; fixed in 1:9.0.120-1.el8_10; e.g. CVE-2024-52316, CVE-2024-54677); atinject 0:1-28.20100611svn86.module+el8+2452+b359bfcd (2 CVEs, critical; fixed in 0:1.0.5-5.module+el8.10.0+21301+657f54a3; e.g. CVE-2022-29599, CVE-2025-67030); java-17-openjdk-headless-slowdebug 1:17.0.8.0.7-2.el8 (50 CVEs, high; fixed in 1:17.0.20.0.8-1.1.el8; e.g. CVE-2023-48161, CVE-2024-21208); and 1 more. 33 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 353 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall inbound policy\", \"finding\": \"The device reports allow (no firewall), but the baseline requires deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"Found: none running. Required: antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: within limit\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: rschmit\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout: meets the baseline\", \"finding\": \"The device reports within limit, which meets the requirement (<= 300 seconds).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect 5393, 20 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Compliant\", \"title\": \"Malware protection signatures matches the required setting\", \"finding\": \"Found: XProtect 5393, 20 days old. This satisfies the baseline requirement of <= 30 days old.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: FullyEncrypted (100%)\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Encryption is compliant\", \"finding\": \"Encryption is set to FullyEncrypted (100%), in line with the required fully encrypted.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Network Time: Off\nREQUIRED: network time on\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation is non-compliant\", \"finding\": \"Found: Network Time: Off. Required: network time on. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"sudo systemsetup -setusingnetworktime on.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates is non-compliant\", \"finding\": \"The user reports manual, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: auditd running, journal persistent\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging: meets the baseline\", \"finding\": \"Found: auditd running, journal persistent. This satisfies the baseline requirement of auditd running.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Alpaca (Ollama client)\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: Alpaca (Ollama client). Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Found: partial. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Screen timeout 10 minutes\nREQUIRED: <= 300 seconds\nDETAIL: The phone stays unlocked for 10 minutes when not in use.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen timeout\", \"finding\": \"Screen timeout is Screen timeout 10 minutes instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no passcode\nREQUIRED: passcode set\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Passcode is non-compliant\", \"finding\": \"Found: no passcode. Required: passcode set. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Signed-in user account type on Windows: Signed-in user account type on Windows. Why it matters: Day-to-day work under an administrator account lets any malware or mistake change the whole system. Risk if missing: High - Malware started by the signed-in user runs with full administrator rights. Requirement: The everyday signed-in user is a Standard account (not an administrator), unless listed in APPROVED_ADMINS. How to fix on Windows: (1) Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation. How to verify: Re-run the audit; Signed-in user account type should read Standard. Related controls: ISO 27001:2022 A.8.2 (Privileged access rights); ISO 27001:2022 A.5.15 (Access control); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] How to install on Windows: Run the downloaded setup .exe. Windows SmartScreen shows 'Windows protected your PC' because the app is not yet code-signed: click More info, then Run anyway. Click Next, Install and Finish. When the audit starts, click Yes in the User Account Control prompt. The Windows version is a beta.\n[3] How the phone check works: Phones are checked in the browser: open the Devseis page on the iPhone or Android phone and start the phone check. The browser detects what it can (system version, whether a screen lock is set up) and asks the rest with help on where to look in Settings. Every result is labelled detected or self-reported. The report can be saved or shared as a PDF.\n[4] USB storage device access on Windows: USB storage device access on Windows. Why it matters: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives. Risk if missing: Low - The device access setting differs from the organisation's agreed USB policy. Requirement: Whether USB mass storage devices can connect at all equals USB_DEVICE_ACCESS (allow/deny): driver, device-installation policy, USBGuard or MDM restriction. How to fix on Windows: (1) To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4. How to verify: Re-run the audit; USB storage device access should read allow. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.12 (Data leakage prevention); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Microsoft Windows 11 Enterprise 10.0.26100 (windows), audited 2026-10-05; 46 of 55 assessed checks compliant; 9 gaps, 0 pending, 1 not read. Gaps: USB storage device access: deny; USB storage read: deny; Signed-in user account type: Administrator (CONTOSO\\pmartin); Pending security updates: 2 pending; Outdated third-party applications: 1 outdated; Backup configured: none found; Recent backup: no backup; Data processing agreements: partial.\n\nQUESTION: Where do I start?"}, {"role": "assistant", "content": "Your audit found 9 gaps. Start with USB storage device access (found: deny). Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives. To fix it [4]:\n1. To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\n\nThen look at: USB storage read, Signed-in user account type, Pending security updates."}], "meta": {"task": "chat", "os": "windows", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Red Hat Enterprise Linux 8.10 (Ootpa)\nREQUIRED: RHEL 9 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"The device reports Red Hat Enterprise Linux 8.10 (Ootpa), but the baseline requires RHEL 9 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound is non-compliant\", \"finding\": \"Public outbound is allow instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iCloud Backup\", \"finding\": \"iCloud Backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Answer ORG_DPIA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Windows default (automatic)\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates matches the required setting\", \"finding\": \"Automatic security updates is set to Windows default (automatic), in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-10 Cumulative Update for Windows 10 Version 22H2 for x64-based Systems (KB5042527), Security Intelligence Update for Microsoft Defender Antivirus - KB2267602 (Version 1.459.0).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Found: 2 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Password history is 0 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Jan\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Jan. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: mkhan. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 16 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 16\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"Outdated third-party applications is 16 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption protection: does not meet the baseline\", \"finding\": \"The device reports not encrypted, but the baseline requires active, strong cipher. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2026-02 (8 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2026-02 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Android security update\", \"finding\": \"The user reports security patch 2026-02 (8 months old), but the baseline requires <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.7 (ios)\nHOSTNAME: OPS-iPhone-03\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=16, Non-Compliant=2, NotApplicable=3, Pending=8, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 30 seconds\n- mobile.passcode_strength | Passcode strength | Compliant | 6-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store and employer-managed apps\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.7 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 15 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | When Unlocked\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\nSCORE (computed): compliant=16, assessed=18, percent=89\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | iOS security updates | High | iOS 26.7 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 15 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"OPS-iPhone-03 - iOS 26.7\", \"score\": {\"compliant\": 16, \"assessed\": 18, \"percent\": 89}, \"counts\": {\"Compliant\": 16, \"Non-Compliant\": 2, \"NotApplicable\": 3, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.7 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"15 critical/high\"}], \"summary\": \"OPS-iPhone-03 meets 16 of 18 assessed checks (89%). The most serious gaps are iOS security updates and Known vulnerabilities in iOS. 8 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_faillock not enabled\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"Found: pam_faillock not enabled. Required: 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\", \"Set deny = 5 in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength is non-compliant\", \"finding\": \"The user reports 4-digit code, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 12 (android)\nHOSTNAME: dev-galaxy-48\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=8, Non-Compliant=11, NotApplicable=2, Pending=9, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 30 seconds\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | 4-digit PIN\n- mobile.jailbreak_root | Not rooted | Non-Compliant | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 12 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2025-12 (10 months old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 12\n- gdpr.time_sync | Date & time set automatically | Pending | not answered\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- gdpr.auto_updates: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=8, assessed=19, percent=42\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not rooted | High | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- vuln.pending_updates | Android security update | High | security patch 2025-12 (10 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.os_supported | Supported operating system version | High | Android 12\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone."}, {"role": "assistant", "content": "{\"device\": \"dev-galaxy-48 - Android 12\", \"score\": {\"compliant\": 8, \"assessed\": 19, \"percent\": 42}, \"counts\": {\"Compliant\": 8, \"Non-Compliant\": 11, \"NotApplicable\": 2, \"Pending\": 9, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not rooted\", \"risk\": \"High\", \"found\": \"rooted or bootloader unlocked\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2025-12 (10 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 12\"}], \"summary\": \"dev-galaxy-48 meets 8 of 19 assessed checks (42%). The most serious gaps are Not rooted, Install unknown apps blocked and Android security update. 9 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not rooted: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\", \"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: GPT4All\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (linux):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"AI software inventory is set to GPT4All, in line with the required inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Network Time: Off\nREQUIRED: network time on\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Time synchronisation\", \"finding\": \"Found: Network Time: Off. Required: network time on. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"sudo systemsetup -setusingnetworktime on.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: rschmit\nREQUIRED: password required\nDETAIL: Users checked: rschmit\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"Found: not required for: rschmit. Required: password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Found: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security intelligence [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 3 days old (version 1.432.96.0)\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (windows):\n- Run Update-MpSignature and check that the device can reach Windows Update or your update server.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Compliant\", \"title\": \"Security intelligence matches the required setting\", \"finding\": \"Security intelligence is set to 3 days old (version 1.432.96.0), in line with the required <= 3 days old.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 18.7 (update to 18.7.10 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 18.7.10 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates: does not meet the baseline\", \"finding\": \"The device reports iOS 18.7 (update to 18.7.10 available), but the baseline requires latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (windows):\n- Enable real-time protection in Microsoft Defender (on-access protection follows it).\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access protection: does not meet the baseline\", \"finding\": \"The device reports off, but the baseline requires on. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Enable real-time protection in Microsoft Defender (on-access protection follows it).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Data protection impact assessment is partial instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after 900 seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"Found: password after 900 seconds. Required: password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic system updates\", \"finding\": \"Automatic system updates is manual instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nHOSTNAME: legal-dev-71\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=40, Non-Compliant=12, NotApplicable=2, Pending=2, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 99999\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | minclass=0, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 3 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (pmartin)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 21 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Debian GNU/Linux 12 (bookworm)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Claude Code\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.expiration: Low\n- password.complexity: Medium\n- malware.on_access: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- aiact.unapproved: Medium\n- org.dpia: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=40, assessed=52, percent=77\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\n- gdpr.backup_configured | Backup configured | High | none\n- password.min_length | Minimum password length | Medium | 6\n- password.complexity | Password complexity | Medium | minclass=0, required classes=0\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\n- firewall.inbound: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\n- gdpr.backup_configured: Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup)."}, {"role": "assistant", "content": "{\"device\": \"legal-dev-71 - Debian GNU/Linux 12 (bookworm) 12\", \"score\": {\"compliant\": 40, \"assessed\": 52, \"percent\": 77}, \"counts\": {\"Compliant\": 40, \"Non-Compliant\": 12, \"NotApplicable\": 2, \"Pending\": 2, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"password.min_length\", \"name\": \"Minimum password length\", \"risk\": \"Medium\", \"found\": \"6\"}, {\"check_id\": \"password.complexity\", \"name\": \"Password complexity\", \"risk\": \"Medium\", \"found\": \"minclass=0, required classes=0\"}], \"summary\": \"legal-dev-71 meets 40 of 52 assessed checks (77%). The most serious gaps are Firewall, Firewall inbound policy and Backup configured. 2 items are pending.\", \"next_steps\": [\"Firewall: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\", \"Firewall inbound policy: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\", \"Backup configured: Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Find My Device\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Jan, GPT4All\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The device reports Jan, GPT4All, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"Found: no. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised: does not meet the baseline\", \"finding\": \"The user reports usage & diagnostics on, but the baseline requires usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"The device reports partial, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"Data protection impact assessment is partial instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"Non-Compliant\", \"title\": \"Tamper protection: does not meet the baseline\", \"finding\": \"Tamper protection is off instead of on. Tamper protection stops malware or users switching off Defender.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Attackers commonly disable antivirus before deploying ransomware.\"}, \"recommendation\": [\"Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Tamper protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: itadmin\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Compliant\", \"title\": \"Approved IT administrator is compliant\", \"finding\": \"The device reports itadmin, which meets the requirement (itadmin).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"Found: not set. Required: >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nHOSTNAME: ops-lt-85\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=38, Non-Compliant=11, NotApplicable=2, Pending=4, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Non-Compliant | PASS_MIN_DAYS 0\n- password.complexity | Password complexity | Compliant | minclass=0, required classes=4\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Compliant | 0 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | firewalld active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | mkhan\n- user.account_type | Signed-in user account type | Compliant | Standard (ewagner)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 3 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 11 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 1989 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Compliant | 27 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Red Hat Enterprise Linux 9.6 (Plow)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Non-Compliant | NTP service no, synchronised no\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Error | not read\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Hugging Face models\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Hugging Face models\n- aiact.risk_classification | AI system risk classification | Pending | Hugging Face models\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_age: Low\n- malware.signature_updates: Medium\n- malware.on_access: Medium\n- admin.unauthorized: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- aiact.unapproved: Medium\n- org.ropa: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=38, assessed=49, percent=78\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee administrator accounts | High | mkhan\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 1989 critical/high\n- gdpr.backup_configured | Backup configured | High | none\n- org.ropa | Record of processing activities | High | partial\n- malware.signature_updates | Malware signature updates | Medium | freshclam not running\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.backup_configured: Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup)."}, {"role": "assistant", "content": "{\"device\": \"ops-lt-85 - Red Hat Enterprise Linux 9.6 (Plow) 9.6\", \"score\": {\"compliant\": 38, \"assessed\": 49, \"percent\": 78}, \"counts\": {\"Compliant\": 38, \"Non-Compliant\": 11, \"NotApplicable\": 2, \"Pending\": 4, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"mkhan\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"1989 critical/high\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"malware.signature_updates\", \"name\": \"Malware signature updates\", \"risk\": \"Medium\", \"found\": \"freshclam not running\"}], \"summary\": \"ops-lt-85 meets 38 of 49 assessed checks (78%). The most serious gaps are Unauthorized employee administrator accounts, Known vulnerabilities in installed software and Backup configured. 1 checks could not be read. 4 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Backup configured: Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: LM Studio, Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: LM Studio, Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no signature database\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures is non-compliant\", \"finding\": \"The device reports no signature database, but the baseline requires <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google Play Protect\", \"finding\": \"The user reports Play Protect scanning off, but the baseline requires on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: passcode set\nREQUIRED: passcode set\nDETAIL: The browser reports that a device unlock method (passcode, PIN, Face ID or fingerprint) is set up.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Passcode is compliant\", \"finding\": \"Passcode is set to passcode set, in line with the required passcode set.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Hugging Face models\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: Hugging Face models. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 27.0 (ios)\nAI TOOLS FOUND: Gemini\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Gemini.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"On-access malware protection is not running instead of on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: ChatGPT. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Found: 12. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 90; accounts: jdoe (99999)\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"The device reports PASS_MAX_DAYS 90; accounts: jdoe (99999), but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: <= 90 days old\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Pending\", \"title\": \"Android security update is pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: no guest login\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled matches the required setting\", \"finding\": \"The device reports no guest login, which meets the requirement (no guest login).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Minimum password length is 12 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Screen timeout 30 seconds\nREQUIRED: <= 300 seconds\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen timeout is compliant\", \"finding\": \"Screen timeout is set to Screen timeout 30 seconds, in line with the required <= 300 seconds. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone is non-compliant\", \"finding\": \"Find My iPhone is off instead of on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security intelligence [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 24 days old (version 1.430.686.0)\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (windows):\n- Run Update-MpSignature and check that the device can reach Windows Update or your update server.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Security intelligence is non-compliant\", \"finding\": \"Security intelligence is 24 days old (version 1.430.686.0) instead of <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Run Update-MpSignature and check that the device can reach Windows Update or your update server.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification does not apply here\", \"finding\": \"This check does not apply: no AI tools found. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: screen lock set\nREQUIRED: PIN, password or pattern\nDETAIL: The browser reports that a device unlock method (passcode, PIN, Face ID or fingerprint) is set up.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock: meets the baseline\", \"finding\": \"The device reports screen lock set, which meets the requirement (PIN, password or pattern).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iPadOS 18.7.9 (update to 18.7.10 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iPadOS 18.7.10 or later in Settings › General › Software Update.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iOS security updates\", \"finding\": \"Found: iPadOS 18.7.9 (update to 18.7.10 available). Required: latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-09 Cumulative Update for Windows 10 Version 22H2 for x64-based Systems (KB5048830).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates: does not meet the baseline\", \"finding\": \"Pending security updates is 1 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Sensitive notifications hidden when locked: does not meet the baseline\", \"finding\": \"The user reports all content shown, but the baseline requires sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.7 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iOS security updates\", \"finding\": \"iOS security updates is iOS 26.7 (update to 26.7.1 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nHOSTNAME: hr-lt-36\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=18, Non-Compliant=31, NotApplicable=4, Pending=2, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 99999\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | pam_pwquality not enabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Non-Compliant | no signature database\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Non-Compliant | off for: lmuller\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: lmuller\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | rschmit, mkhan\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (lmuller)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 3 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 15 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 95 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 168 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Debian GNU/Linux 13 (trixie)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Timeshift\n- gdpr.backup_recent | Recent backup | Non-Compliant | 37 days ago\n- gdpr.removable_encryption | Removable media encryption | Error | not read\n- gdpr.guest_account | Guest account disabled | Non-Compliant | LightDM guest session allowed\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | teamviewer\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | popularity-contest on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.expiration: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- malware.antivirus: High\n- malware.service: High\n- malware.signature_updates: Medium\n- malware.signatures: High\n- malware.on_access: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.backup_recent: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.dpia: Medium\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=18, assessed=49, percent=37\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.service | ClamAV malware protection service | High | not running\n- malware.signatures | Malware signatures | High | no signature database\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now."}, {"role": "assistant", "content": "{\"device\": \"hr-lt-36 - Debian GNU/Linux 13 (trixie) 13\", \"score\": {\"compliant\": 18, \"assessed\": 49, \"percent\": 37}, \"counts\": {\"Compliant\": 18, \"Non-Compliant\": 31, \"NotApplicable\": 4, \"Pending\": 2, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"no signature database\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}], \"summary\": \"hr-lt-36 meets 18 of 49 assessed checks (37%). The most serious gaps are Antivirus, ClamAV malware protection service and Malware signatures. 1 checks could not be read. 2 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Compliant\", \"title\": \"AI transparency to users: meets the baseline\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 50\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (mkhan)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"The device reports Administrator (mkhan), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Block or allow external storage with your MDM (Restrictions / device control) or an endpoint DLP product.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access: meets the baseline\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAI TOOLS FOUND: Jan, GPT4All\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Jan: Local model runner; provider Open source (runs locally); data: On the device unless a cloud model is connected; default risk minimal; GDPR: Check whether remote API providers are configured; if so, the cloud provider rules apply.\n- GPT4All: Local model runner; provider Nomic AI (runs locally); data: On the device unless optional data sharing is enabled; default risk minimal; GDPR: Keep the optional data lake sharing switched off.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Jan\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Check whether remote API providers are configured; if so, the cloud provider rules apply.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"GPT4All\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Keep the optional data lake sharing switched off.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Jan, GPT4All.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection is non-compliant\", \"finding\": \"Found: not running. Required: on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 11 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 11\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"Outdated third-party applications is 11 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Cursor\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Current state: Cursor. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access is compliant\", \"finding\": \"USB storage device access is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: File History (jdoe)\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Compliant\", \"title\": \"Backup configured: meets the baseline\", \"finding\": \"The device reports File History (jdoe), which meets the requirement (backup configured).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Passcode strength\", \"finding\": \"Found: 4-digit code. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pattern\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: Pattern locks are easy to guess from screen smudges or by watching.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength: does not meet the baseline\", \"finding\": \"Found: pattern. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 20 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 18; flatpaks: org.libreoffice.LibreOffice, org.mozilla.firefox\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Outdated third-party applications is 20 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off (no passcode)\nREQUIRED: on\nDETAIL: Without a passcode, iOS data protection is not active.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Data protection (encryption) is non-compliant\", \"finding\": \"Found: off (no passcode). Required: on. Full-disk encryption keeps data unreadable if the device is lost or stolen. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; too long or off for: FABRIKAM\\jdoe (600s, active=0)\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"The device reports no machine inactivity limit; too long or off for: FABRIKAM\\\\jdoe (600s, active=0), but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: Claude Code. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: vnc\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Found: vnc. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: on-access scanning\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Error\", \"title\": \"On-access malware protection could not be read\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 66 days old\nREQUIRED: <= 3 days old\nDETAIL: /var/lib/clamav/daily.cld\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures is non-compliant\", \"finding\": \"Malware signatures is 66 days old instead of <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: LM Studio, Ollama\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Found: LM Studio, Ollama. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"The device reports 0, but the baseline requires >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\", \"For a stand-alone device: net accounts /uniquepw:24\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The device reports Claude Code, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Meta AI (AI tool): Meta AI: Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps), provider Meta (US), data location: Provider cloud outside the EEA. Default EU AI Act risk class for general office use: minimal. GDPR: Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[2] GDPR Art. 28 Processor: GDPR Art. 28 (Processor): Use only processors that give sufficient guarantees, under a written contract with required terms. An auditor looks for: A data processing agreement with each processor. On endpoints: Cloud AI tools used with work data are processors and need an agreement.\n[3] Gemini (AI tool): Gemini: Cloud AI assistant, provider Google (US), data location: Provider cloud. Default EU AI Act risk class for general office use: minimal. GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[4] Cursor (AI tool): Cursor: AI code editor, provider Anysphere (US), data location: Code sent to the provider and its model providers unless privacy mode is enforced. Default EU AI Act risk class for general office use: minimal. GDPR: Enforce privacy mode through a business plan and sign a data processing agreement. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer (x86-64), Firefox 143; detected by the browser, security settings not checked.\n\nQUESTION: Can I use Cursor with customer data under GDPR?"}, {"role": "assistant", "content": "Cursor is a ai code editor from Anysphere (US); data goes to: code sent to the provider and its model providers unless privacy mode is enforced [4]. For general office use its default EU AI Act risk class is minimal. GDPR: Enforce privacy mode through a business plan and sign a data processing agreement.\n\nWhether it is allowed depends on your organisation's approved AI tools list and on what you use it for: some purposes (for example screening job applicants) make any AI tool high-risk."}], "meta": {"task": "chat", "os": "any", "status": "tool", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 10 Pro build 19045\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"Found: Microsoft Windows 10 Pro build 19045. Required: build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: automatic app updates on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Pending\", \"title\": \"App updates: decision pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Compliant\", \"title\": \"Private outbound is compliant\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Compliant\", \"title\": \"AI transparency to users: meets the baseline\", \"finding\": \"AI transparency to users is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 50\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: GPT4All\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: GPT4All. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 27.0 (ios)\nHOSTNAME: HR-iPhone-09\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=17, Non-Compliant=1, NotApplicable=3, Pending=8, Error=0\nEVIDENCE: 2 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 30 seconds\n- mobile.passcode_strength | Passcode strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store and employer-managed apps\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 27.0\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | When Unlocked\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- vuln.outdated_apps: Medium\nSCORE (computed): compliant=17, assessed=18, percent=94\nTOP GAPS (computed, highest risk first):\n- vuln.outdated_apps | App updates | Medium | automatic app updates off\nFIX GUIDANCE (highest-risk gaps):\n- vuln.outdated_apps: Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps."}, {"role": "assistant", "content": "{\"device\": \"HR-iPhone-09 - iOS 27.0\", \"score\": {\"compliant\": 17, \"assessed\": 18, \"percent\": 94}, \"counts\": {\"Compliant\": 17, \"Non-Compliant\": 1, \"NotApplicable\": 3, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.outdated_apps\", \"name\": \"App updates\", \"risk\": \"Medium\", \"found\": \"automatic app updates off\"}], \"summary\": \"HR-iPhone-09 meets 17 of 18 assessed checks (94%). There are 1 gaps, none of them rated high risk. 8 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"App updates: Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall outbound policy\", \"finding\": \"Found: no firewall. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 228 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-02-15 01:48:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security updates installed recently\", \"finding\": \"The device reports 228 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 53 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 25 applications inventoried, 13 recognised (23 common applications covered). Microsoft Teams 1.6.00.26463 (7 CVEs, high, actively exploited; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330); Visual Studio Code 1.100.1 (29 CVEs, critical; fixed in 1.136.2; e.g. CVE-2025-55319, CVE-2025-64660); VLC 2.2.1 (17 CVEs, critical; fixed in 3.0.20; e.g. CVE-2016-5108, CVE-2017-10699). 23 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 53 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Data protection impact assessment is no instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAI TOOLS FOUND: Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 314 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 21 applications inventoried, 10 recognised (23 common applications covered). Adobe Acrobat Reader 25.001.20432 (49 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-43550, CVE-2025-43573); Firefox 152.3 (201 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100768, CVE-2026-100779); Google Chrome 154.0.8037.97 (58 CVEs, critical; fixed in 155.0.8059.39; e.g. CVE-2026-106220, CVE-2026-106221); Microsoft Teams 2.10.1 (6 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330). 138 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 314 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength: does not meet the baseline\", \"finding\": \"The user reports 4-digit code, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Found: 6. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Jan\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Unapproved AI tools (shadow AI) is Jan instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"The device reports none running, but the baseline requires antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: EncryptionInProgress (42%)\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"The device reports EncryptionInProgress (42%), but the baseline requires fully encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: GPT4All\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (linux):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"AI software inventory is set to GPT4All, in line with the required inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Ventura 13.8.9-24H501.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"The device reports 1 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery: does not meet the baseline\", \"finding\": \"Disk encryption recovery is 1 key slots, 0 unlock tokens instead of recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, XtsAes128\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker protection is non-compliant\", \"finding\": \"The device reports protection Off, XtsAes128, but the baseline requires protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-09 Cumulative Update for Windows 11 Version 22H2 for x64-based Systems (KB5041580).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Pending security updates is 1 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: not jailbroken (user confirmed)\nREQUIRED: not jailbroken\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Compliant\", \"title\": \"Not jailbroken matches the required setting\", \"finding\": \"Not jailbroken is set to not jailbroken (user confirmed), in line with the required not jailbroken. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"The device reports no lockout configured, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutwindow:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Compliant\", \"title\": \"Unauthorized employee administrator accounts: meets the baseline\", \"finding\": \"Found: none. This satisfies the baseline requirement of none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"Screen lock timeout is no machine inactivity limit; screen saver off instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: not checked\nREQUIRED: <= 0 outdated\nDETAIL: Homebrew is not installed, so there is no local source of current application versions.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (macos):\n- Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\nverify: Re-run the audit; outdated applications should be 0 or fewer.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"NotApplicable\", \"title\": \"Outdated third-party applications: not assessed on this device\", \"finding\": \"This check does not apply: Homebrew is not installed, so there is no local source of current application versions.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 251 source packages checked against ubuntu:26.04 advisories. 9 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Compliant\", \"title\": \"Known vulnerabilities in installed software is compliant\", \"finding\": \"Known vulnerabilities in installed software is set to 0 critical/high, in line with the required <= 0 critical/high.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption protection: does not meet the baseline\", \"finding\": \"Disk encryption protection is not encrypted instead of active, strong cipher. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 4 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-05 Cumulative Update for Windows 11 (2026-10-01)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Compliant\", \"title\": \"Security updates installed recently: meets the baseline\", \"finding\": \"Security updates installed recently is set to 4 days ago, in line with the required <= 30 days ago.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answer ORG_AI_TRANSPARENCY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Minimum password length is 10 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: not jailbroken (user confirmed)\nREQUIRED: not jailbroken\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Compliant\", \"title\": \"Not jailbroken is compliant\", \"finding\": \"Found: not jailbroken (user confirmed). This satisfies the baseline requirement of not jailbroken. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: automatic app updates on\nREQUIRED: automatic app updates on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Compliant\", \"title\": \"App updates is compliant\", \"finding\": \"App updates is set to automatic app updates on, in line with the required automatic app updates on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup: does not meet the baseline\", \"finding\": \"Found: no backup. Required: backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period matches the required setting\", \"finding\": \"The device reports 30 minutes, which meets the requirement (>= 15 minutes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: fully encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"Disk encryption protection matches the required setting\", \"finding\": \"The device reports fully encrypted, which meets the requirement (fully encrypted).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 15 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Compliant\", \"title\": \"Account lockout duration matches the required setting\", \"finding\": \"Found: 15 minutes. This satisfies the baseline requirement of >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, None\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker protection\", \"finding\": \"The device reports protection Off, None, but the baseline requires protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Tahoe 26.0.2-22H363, Command Line Tools for Xcode-16.4.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Found: 2 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security intelligence [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 19 days old (version 1.434.479.0)\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (windows):\n- Run Update-MpSignature and check that the device can reach Windows Update or your update server.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security intelligence\", \"finding\": \"Found: 19 days old (version 1.434.479.0). Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Run Update-MpSignature and check that the device can reach Windows Update or your update server.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 26.0.1 (macos)\nAI TOOLS FOUND: ChatGPT\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 180 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration is compliant\", \"finding\": \"The device reports 180 days, which meets the requirement (1-365 days).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"Account lockout threshold is 10 instead of 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"The device reports 1 key slots, 0 unlock tokens, but the baseline requires recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: enabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Compliant\", \"title\": \"Password complexity: meets the baseline\", \"finding\": \"The device reports enabled, which meets the requirement (enabled).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Google backup: does not meet the baseline\", \"finding\": \"Google backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Homebrew packages checked; App Store apps are not covered.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (macos):\n- Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Compliant\", \"title\": \"Outdated third-party applications: meets the baseline\", \"finding\": \"The device reports 0 outdated, which meets the requirement (<= 0 outdated).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: sali (0s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: sali, asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: too long for: sali (0s). Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 27.0 (ios)\nHOSTNAME: FIN-iPhone-08\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=15, Non-Compliant=2, NotApplicable=1, Pending=11, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 5 minutes\n- mobile.passcode_strength | Passcode strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 27.0\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Pending | not answered\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | When Unlocked\n- aiact.inventory | AI software inventory | Compliant | Gemini\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Gemini\n- aiact.risk_classification | AI system risk classification | Pending | Gemini\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.app_sources: High\n- aiact.unapproved: Medium\nSCORE (computed): compliant=15, assessed=17, percent=88\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | Gemini\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it."}, {"role": "assistant", "content": "{\"device\": \"FIN-iPhone-08 - iOS 27.0\", \"score\": {\"compliant\": 15, \"assessed\": 17, \"percent\": 88}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 2, \"NotApplicable\": 1, \"Pending\": 11, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"Gemini\"}], \"summary\": \"FIN-iPhone-08 meets 15 of 17 assessed checks (88%). The most serious gaps are Apps only from the App Store. 11 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Unapproved AI tools (shadow AI) is ChatGPT instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 12 (android)\nHOSTNAME: sales-oneplus-52\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=12, Non-Compliant=8, NotApplicable=2, Pending=8, Error=0\nEVIDENCE: 2 detected by the browser, 20 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Non-Compliant | no screen lock (Swipe/None)\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 2 minutes\n- mobile.passcode_strength | Screen lock strength | Compliant | 6-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 12 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-10 (0 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 12\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.enabled: Medium\n- mobile.find_my_device: Medium\n- vuln.outdated_apps: Medium\n- gdpr.os_supported: High\n- gdpr.time_sync: Low\n- gdpr.lock_screen_previews: Low\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=12, assessed=20, percent=60\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- gdpr.os_supported | Supported operating system version | High | Android 12\n- screen_lock.enabled | Screen lock | Medium | no screen lock (Swipe/None)\n- mobile.find_my_device | Find My Device | Medium | off\n- vuln.outdated_apps | App updates | Medium | automatic app updates off\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- gdpr.os_supported: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\n- screen_lock.enabled: Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password."}, {"role": "assistant", "content": "{\"device\": \"sales-oneplus-52 - Android 12\", \"score\": {\"compliant\": 12, \"assessed\": 20, \"percent\": 60}, \"counts\": {\"Compliant\": 12, \"Non-Compliant\": 8, \"NotApplicable\": 2, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 12\"}, {\"check_id\": \"screen_lock.enabled\", \"name\": \"Screen lock\", \"risk\": \"Medium\", \"found\": \"no screen lock (Swipe/None)\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My Device\", \"risk\": \"Medium\", \"found\": \"off\"}, {\"check_id\": \"vuln.outdated_apps\", \"name\": \"App updates\", \"risk\": \"Medium\", \"found\": \"automatic app updates off\"}], \"summary\": \"sales-oneplus-52 meets 12 of 20 assessed checks (60%). The most serious gaps are Google Play Protect and Supported operating system version. 8 items are pending. 20 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Supported operating system version: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\", \"Screen lock: Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Record of processing activities is partial instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, XtsAes128\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker protection\", \"finding\": \"BitLocker protection is protection Off, XtsAes128 instead of protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"Found: partial. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iPadOS 26.5 (ios)\nHOSTNAME: MKT-iPad-22\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=14, Non-Compliant=4, NotApplicable=2, Pending=9, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 5 minutes\n- mobile.passcode_strength | Passcode strength | Compliant | 7+ digit numeric code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Pending | not answered\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iPadOS 26.5 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 106 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | off\n- gdpr.os_supported | Supported operating system version | Compliant | iPadOS 26.5\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | backed up to a computer\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | ChatGPT extension off\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.telemetry: Low\nSCORE (computed): compliant=14, assessed=18, percent=78\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | iOS security updates | High | iPadOS 26.5 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 106 critical/high\n- gdpr.auto_updates | Automatic iOS updates | High | off\n- gdpr.telemetry | Analytics sharing minimised | Low | analytics sharing on\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.auto_updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files."}, {"role": "assistant", "content": "{\"device\": \"MKT-iPad-22 - iPadOS 26.5\", \"score\": {\"compliant\": 14, \"assessed\": 18, \"percent\": 78}, \"counts\": {\"Compliant\": 14, \"Non-Compliant\": 4, \"NotApplicable\": 2, \"Pending\": 9, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iPadOS 26.5 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"106 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"gdpr.telemetry\", \"name\": \"Analytics sharing minimised\", \"risk\": \"Low\", \"found\": \"analytics sharing on\"}], \"summary\": \"MKT-iPad-22 meets 14 of 18 assessed checks (78%). The most serious gaps are iOS security updates, Known vulnerabilities in iOS and Automatic iOS updates. 9 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Automatic iOS updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (linux):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"Found: no AI tools found. This satisfies the baseline requirement of inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption protection\", \"finding\": \"Found: not encrypted. Required: fully encrypted. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is set to none, in line with the required only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Grammarly\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: ChatGPT, Grammarly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jailbroken\nREQUIRED: not jailbroken\nDETAIL: iOS security protections are bypassed on a jailbroken phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not jailbroken: does not meet the baseline\", \"finding\": \"Not jailbroken is jailbroken instead of not jailbroken. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 50 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 16 applications inventoried, 11 recognised (31 common applications covered). 7-Zip 24.07 (x64) 24.07 (8 CVEs, high, actively exploited; fixed in 26.02; e.g. CVE-2025-0411, CVE-2025-53816); Adobe Acrobat Reader (64-bit) 26.001.21651 (35 CVEs, high; fixed in 26.002.21901; e.g. CVE-2026-47911, CVE-2026-47912); Microsoft Edge 152.0.4191.52 (5 CVEs, high; fixed in 153.0.4234.46; e.g. CVE-2026-69486, CVE-2026-72984); Microsoft Teams 2020-10-29 (2 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2025-49737, CVE-2025-53783). 30 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 50 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Compliant\", \"title\": \"Domain firewall: meets the baseline\", \"finding\": \"Domain firewall is set to on, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in ArqAgent.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Current state: date not readable. Confirm the last backup date in ArqAgent.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: Claude Code. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 42 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 13 applications inventoried, 10 recognised (23 common applications covered). Safari 18.6 (28 CVEs, critical, actively exploited; fixed in 26.6.1; e.g. CVE-2025-14174, CVE-2025-43342); Visual Studio Code 1.132.1 (10 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-70334, CVE-2026-78461); iTerm 3.4.20 (3 CVEs, critical; e.g. CVE-2023-46321, CVE-2023-46322); zoom.us 7.0.0 (1 CVE, high; fixed in 7.0.5; e.g. CVE-2026-53410). 95 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 42 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Found: partial. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold: does not meet the baseline\", \"finding\": \"Found: not set. Required: 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxFailedAttempts = 5.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 6 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-04 Cumulative Update for Windows 11 (2026-09-28)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Compliant\", \"title\": \"Security updates installed recently is compliant\", \"finding\": \"Security updates installed recently is set to 6 days ago, in line with the required <= 30 days ago.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Gemini, Grammarly\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: ChatGPT, Gemini, Grammarly. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 22 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-06 Cumulative Update for Windows 11 (2026-09-11)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Compliant\", \"title\": \"Security updates installed recently is compliant\", \"finding\": \"Found: 22 days ago. This satisfies the baseline requirement of <= 30 days ago.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: not jailbroken (user confirmed)\nREQUIRED: not jailbroken\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Compliant\", \"title\": \"Not jailbroken: meets the baseline\", \"finding\": \"Found: not jailbroken (user confirmed). This satisfies the baseline requirement of not jailbroken. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: download only\nREQUIRED: download and install on\nDETAIL: Updates wait for the user to install them.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic iOS updates\", \"finding\": \"Found: download only. Required: download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Hugging Face models, Ollama, Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Unapproved AI tools (shadow AI) is Hugging Face models, Ollama, Claude Code instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: automatic updates turned off by policy\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic security updates\", \"finding\": \"Found: automatic updates turned off by policy. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Find My iPhone\", \"finding\": \"Find My iPhone is off instead of on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Found: partial. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Found: not configured. Required: required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (linux):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not assessed on this device\", \"finding\": \"This check does not apply: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAI TOOLS FOUND: Ollama, Hugging Face models\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local AI service listening on port 11434. Local models: /home/jdoe/.ollama/models (4G). Local models: /home/jdoe/.cache/huggingface/hub (14G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Ollama, Hugging Face models.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone is non-compliant\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: XProtect only\nREQUIRED: on-access scanning\nDETAIL: XProtect scans at launch and download; macOS has no built-in on-access scanning.\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (macos):\n- Not provided by macOS itself; deploy a managed endpoint protection product if required.\nverify: Re-run the audit; On-access malware protection should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"NotApplicable\", \"title\": \"On-access malware protection does not apply here\", \"finding\": \"This check does not apply: XProtect scans at launch and download; macOS has no built-in on-access scanning.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no signature database\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures is non-compliant\", \"finding\": \"Found: no signature database. Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"Found: 12. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (ewagner)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"Signed-in user account type is Administrator (ewagner) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 13.7.8 (macos)\nHOSTNAME: DEV-MacBook-20\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=23, Non-Compliant=26, NotApplicable=4, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Non-Compliant | 5\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Non-Compliant | not set\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout\n- lockout.observation | Lockout observation period | Non-Compliant | not set\n- malware.antivirus | Antivirus | Compliant | XProtect 5220\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Non-Compliant | XProtect 5276, 104 days old\n- malware.signature_updates | Malware security updates | Non-Compliant | ConfigDataInstall=0, CriticalUpdateInstall=1\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode on\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 240 seconds\n- screen_lock.enabled | Screen lock | Non-Compliant | off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after off seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | lmuller, mkhan\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (sali)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 120 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Compliant | 27 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | macOS 13.7.8\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Claude Code\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.expiration: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.signatures: High\n- malware.signature_updates: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.enabled: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.processor_agreements: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=23, assessed=49, percent=47\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Malware protection signatures | High | XProtect 5276, 104 days old\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode on\n- admin.unauthorized | Unauthorized employee administrator accounts | High | lmuller, mkhan\n- user.account_type | Signed-in user account type | High | Administrator (sali)\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on."}, {"role": "assistant", "content": "{\"device\": \"DEV-MacBook-20 - macOS 13.7.8\", \"score\": {\"compliant\": 23, \"assessed\": 49, \"percent\": 47}, \"counts\": {\"Compliant\": 23, \"Non-Compliant\": 26, \"NotApplicable\": 4, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Malware protection signatures\", \"risk\": \"High\", \"found\": \"XProtect 5276, 104 days old\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode on\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"lmuller, mkhan\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (sali)\"}], \"summary\": \"DEV-MacBook-20 meets 23 of 49 assessed checks (47%). The most serious gaps are Malware protection signatures, Firewall and Firewall inbound policy. 3 items are pending.\", \"next_steps\": [\"Malware protection signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\", \"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Current state: Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock matches the required setting\", \"finding\": \"Screen lock is set to on, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: automatic app updates on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Pending\", \"title\": \"App updates: needs a follow-up step\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: unattended-upgrades installed=1, enabled=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates: does not meet the baseline\", \"finding\": \"Automatic security updates is unattended-upgrades installed=1, enabled=0 instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 75 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-07-22 11:02:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently is non-compliant\", \"finding\": \"Security updates installed recently is 75 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode on\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"Firewall inbound policy is firewall off, stealth mode on instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Found: none. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 pending\nREQUIRED: <= 0 pending\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Compliant\", \"title\": \"Pending security updates: meets the baseline\", \"finding\": \"The device reports 0 pending, which meets the requirement (<= 0 pending).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in ArqAgent.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: decision pending\", \"finding\": \"Confirm the last backup date in ArqAgent. Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ewagner\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: ewagner. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Copilot\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"The user reports Microsoft Copilot, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 180 idle, password delay 300\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: 180 idle, password delay 300. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pattern\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: Pattern locks are easy to guess from screen smudges or by watching.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength: does not meet the baseline\", \"finding\": \"Found: pattern. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Real-time protection [malware.realtime] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Real-time protection scans files and processes as they are used.\nrisk: High - Without it, malware is only found by scheduled scans, after it has run.\nfix (windows):\n- Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\nverify: Re-run the audit; Real-time protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.realtime\", \"status\": \"Non-Compliant\", \"title\": \"Real-time protection is non-compliant\", \"finding\": \"The device reports off, but the baseline requires on. Real-time protection scans files and processes as they are used.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without it, malware is only found by scheduled scans, after it has run.\"}, \"recommendation\": [\"Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Real-time protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.realtime", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nHOSTNAME: mkt-ubuntu-79\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=42, Non-Compliant=9, NotApplicable=2, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 12\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=3, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 0 days old\n- malware.on_access | On-access malware protection | Compliant | clamonacc running\n- firewall.enabled | Firewall | Compliant | firewalld active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | tnguyen\n- user.account_type | Signed-in user account type | Compliant | Standard (asmith)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 29 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 22.04.5 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Jan, GPT4All\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Jan, GPT4All\n- aiact.risk_classification | AI system risk classification | Pending | Jan, GPT4All\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- malware.service: High\n- usb.write: Medium\n- admin.unauthorized: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\nSCORE (computed): compliant=42, assessed=51, percent=82\nTOP GAPS (computed, highest risk first):\n- malware.service | ClamAV malware protection service | High | not running\n- admin.unauthorized | Unauthorized employee administrator accounts | High | tnguyen\n- gdpr.backup_configured | Backup configured | High | none\n- password.min_length | Minimum password length | Medium | 12\n- usb.write | USB storage write | Medium | allow\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- gdpr.backup_configured: Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup)."}, {"role": "assistant", "content": "{\"device\": \"mkt-ubuntu-79 - Ubuntu 22.04.5 LTS 22.04\", \"score\": {\"compliant\": 42, \"assessed\": 51, \"percent\": 82}, \"counts\": {\"Compliant\": 42, \"Non-Compliant\": 9, \"NotApplicable\": 2, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"tnguyen\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"password.min_length\", \"name\": \"Minimum password length\", \"risk\": \"Medium\", \"found\": \"12\"}, {\"check_id\": \"usb.write\", \"name\": \"USB storage write\", \"risk\": \"Medium\", \"found\": \"allow\"}], \"summary\": \"mkt-ubuntu-79 meets 42 of 51 assessed checks (82%). The most serious gaps are ClamAV malware protection service, Unauthorized employee administrator accounts and Backup configured. 3 items are pending.\", \"next_steps\": [\"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Backup configured: Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"The device reports mkhan, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: FileVault on\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Encryption matches the required setting\", \"finding\": \"The device reports FileVault on, which meets the requirement (FileVault on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"Found: 10 minutes. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised: does not meet the baseline\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification does not apply here\", \"finding\": \"This check does not apply: no AI tools found. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register is non-compliant\", \"finding\": \"Found: no. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: none; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security event logging\", \"finding\": \"Security event logging is logon: none; account management: none instead of logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: chromeremotedesktop, anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"Remote access services restricted is chromeremotedesktop, anydesk instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iPadOS 26.5\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Chrome 138 (iOS)).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version matches the required setting\", \"finding\": \"Supported operating system version is set to iPadOS 26.5, in line with the required >= iOS 18.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"Found: no. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall\", \"finding\": \"The device reports no active firewall, but the baseline requires active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apple Intelligence external integrations restricted\", \"finding\": \"The user reports ChatGPT extension on, but the baseline requires external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: automatic app updates on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Pending\", \"title\": \"App updates: needs a follow-up step\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: LM Studio\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"The device reports LM Studio, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Found: 12. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"Non-Compliant\", \"title\": \"Tamper protection: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Tamper protection stops malware or users switching off Defender.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Attackers commonly disable antivirus before deploying ransomware.\"}, \"recommendation\": [\"Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Tamper protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: until an administrator unlocks\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Compliant\", \"title\": \"Account lockout duration: meets the baseline\", \"finding\": \"Found: until an administrator unlocks. This satisfies the baseline requirement of >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"Lockout observation period is 10 minutes instead of >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Deja Dup (jdoe)\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Compliant\", \"title\": \"Backup configured matches the required setting\", \"finding\": \"Found: Deja Dup (jdoe). This satisfies the baseline requirement of backup configured.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: XProtect 5272, 38 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Malware protection signatures\", \"finding\": \"Malware protection signatures is XProtect 5272, 38 days old instead of <= 30 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: On-access malware protection\", \"finding\": \"On-access malware protection is not running instead of on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] AI software inventory on macOS: AI software inventory on macOS. Why it matters: Deployers must know which AI systems they use before they can classify and govern them. Risk if missing: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools. Requirement: AI apps, local AI services and local model folders are listed (Compliant = inventory taken). How to fix on macOS: (1) Add the AI tools found on this device to the organisation's AI register, with owner and purpose. How to verify: The AI register lists every tool in the inventory. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); GDPR Art. 30 (Records of processing activities); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[2] Remote access services restricted on macOS: Remote access services restricted on macOS. Why it matters: Remote access services let others reach the device over the network; only approved ones should be on. Risk if missing: High - Unapproved remote access tools are a common entry point for attackers and support scams. Requirement: Only remote access services listed in APPROVED_REMOTE_SERVICES are enabled. How to fix on macOS: (1) Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host). How to verify: Re-run the audit; only approved services (none) should remain. Related controls: ISO 27001:2022 A.8.20 (Networks security); ISO 27001:2022 A.6.7 (Remote working); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] AI system risk classification on macOS: AI system risk classification on macOS. Why it matters: Each AI system must be checked against the banned practices and the high-risk uses in Annex III. Risk if missing: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards. Requirement: Each AI tool found is classified (prohibited / high-risk / limited / minimal). Done by the auditor model: Pending in the collector output, NotApplicable when no AI tool is found. How to fix on macOS: (1) Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose. How to verify: Every AI tool in the inventory has a classification in the AI register. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); EU AI Act Art. 5 (Prohibited AI practices); EU AI Act Art. 6 (Classification rules for high-risk AI systems); EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)).\n[4] Unapproved AI tools (shadow AI) on macOS: Unapproved AI tools (shadow AI) on macOS. Why it matters: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Risk if missing: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations. Requirement: Every AI tool found is listed in APPROVED_AI_TOOLS. How to fix on macOS: (1) Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it. (2) Publish an AI acceptable-use policy so staff know which tools are allowed. (3) Remove unapproved apps and restrict installations through your MDM. How to verify: Re-run the audit; Unapproved AI tools should read none. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 28 (Processor); GDPR Art. 44 (General principle for transfers); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n\nMY AUDIT: none\n\nQUESTION: How can I set up unapproved ai tools (shadow ai) on my macOS?"}, {"role": "assistant", "content": "AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. To fix it on macOS [4]:\n1. Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n2. Publish an AI acceptable-use policy so staff know which tools are allowed.\n3. Remove unapproved apps and restrict installations through your MDM.\n\nTo check it worked: Re-run the audit; Unapproved AI tools should read none."}], "meta": {"task": "chat", "os": "macos", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities is non-compliant\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-H6TXNO4\\akaya\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts is non-compliant\", \"finding\": \"Found: DESKTOP-H6TXNO4\\\\akaya. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, None\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker protection is non-compliant\", \"finding\": \"BitLocker protection is protection Off, None instead of protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.4 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates is non-compliant\", \"finding\": \"iOS security updates is iOS 26.4 (update to 26.7.1 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"Found: disabled. Required: enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall on, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall inbound policy\", \"finding\": \"The device reports firewall on, stealth mode off, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is ChatGPT instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: download only\nREQUIRED: download and install on\nDETAIL: Updates wait for the user to install them.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic iOS updates\", \"finding\": \"Automatic iOS updates is download only instead of download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: EncryptionInProgress (42%)\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Encryption\", \"finding\": \"Found: EncryptionInProgress (42%). Required: fully encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6 outdated\nREQUIRED: <= 0 outdated\nDETAIL: VLC media player VideoLAN.VLC 1.6.630 2.6.997 winget; Notepad++ (64-bit x64) Notepad++.Notepad++ 33.8.555 34.6.905 winget; Mozilla Firefox (x64 en-US) Mozilla.Firefox 20.7.319 20.5.167 winget; Zoom Workplace Zoom.Zoom 71.1.951 71.0.851 winget; Google Chrome Google.Chrome 36.7.719 37.4.103 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"The device reports 6 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"AI transparency to users is partial instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Ollama, Hugging Face models\nREQUIRED: inventory taken\nDETAIL: Local AI service listening on port 11434. Local models: /home/jdoe/.ollama/models (4G). Local models: /home/jdoe/.cache/huggingface/hub (14G).\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (linux):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"Found: Ollama, Hugging Face models. This satisfies the baseline requirement of inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_pwquality not enabled\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"The device reports pam_pwquality not enabled, but the baseline requires 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: allow\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Error\", \"title\": \"Firewall outbound policy: could not be checked\", \"finding\": \"No evidence was collected for this check (Requires root: run the collector with sudo.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Compliant\", \"title\": \"AI register is compliant\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Find My iPhone\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 50 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 47; snaps: snap-store, slack, thunderbird\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"The device reports 50 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"On-access malware protection is not running instead of on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Private outbound: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (rschmit)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"The device reports Administrator (rschmit), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 12 (android)\nHOSTNAME: legal-pixel-31\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=11, Non-Compliant=8, NotApplicable=2, Pending=9, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Non-Compliant | Screen timeout 30 minutes\n- mobile.passcode_strength | Screen lock strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Non-Compliant | on\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 12 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-10 (0 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 12\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Pending | not answered\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.timeout: Medium\n- mobile.usb_debugging: Medium\n- vuln.outdated_apps: Medium\n- gdpr.auto_updates: High\n- gdpr.os_supported: High\n- gdpr.lock_screen_previews: Low\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=11, assessed=19, percent=58\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.os_supported | Supported operating system version | High | Android 12\n- screen_lock.timeout | Screen timeout | Medium | Screen timeout 30 minutes\n- mobile.usb_debugging | USB debugging off | Medium | on\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- gdpr.auto_updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker).\n- gdpr.os_supported: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone."}, {"role": "assistant", "content": "{\"device\": \"legal-pixel-31 - Android 12\", \"score\": {\"compliant\": 11, \"assessed\": 19, \"percent\": 58}, \"counts\": {\"Compliant\": 11, \"Non-Compliant\": 8, \"NotApplicable\": 2, \"Pending\": 9, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 12\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Screen timeout\", \"risk\": \"Medium\", \"found\": \"Screen timeout 30 minutes\"}, {\"check_id\": \"mobile.usb_debugging\", \"name\": \"USB debugging off\", \"risk\": \"Medium\", \"found\": \"on\"}], \"summary\": \"legal-pixel-31 meets 11 of 19 assessed checks (58%). The most serious gaps are Google Play Protect, Automatic system updates and Supported operating system version. 9 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Automatic system updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker).\", \"Supported operating system version: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Never\nREQUIRED: When Unlocked or Never\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Compliant\", \"title\": \"Notification previews hidden when locked: meets the baseline\", \"finding\": \"The user reports Never, which meets the requirement (When Unlocked or Never). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 14 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.7.1 checked against NVD iOS records. iOS 26.7.1 (14 CVEs, critical; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-43688). 32 medium/low.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS: does not meet the baseline\", \"finding\": \"Found: 14 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: NTP service no, synchronised no\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Time synchronisation\", \"finding\": \"Found: NTP service no, synchronised no. Required: time synchronised. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"timedatectl set-ntp true (or enable chronyd).\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Grammarly\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Found: ChatGPT, Grammarly. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 15 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period: meets the baseline\", \"finding\": \"Lockout observation period is set to 15 minutes, in line with the required >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"The device reports partial, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 36 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: linux-image-generic, openssl, libc-bin. Package lists are 0 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Pending security updates is 36 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall on, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall inbound policy\", \"finding\": \"Firewall inbound policy is firewall on, stealth mode off instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Google Play Protect is non-compliant\", \"finding\": \"The user reports Play Protect scanning off, but the baseline requires on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 60 days old\nREQUIRED: <= 3 days old\nDETAIL: /var/lib/clamav/daily.cld\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Malware signatures\", \"finding\": \"Malware signatures is 60 days old instead of <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength: does not meet the baseline\", \"finding\": \"Found: 4-digit PIN. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: alphanumeric\nREQUIRED: >= 6 digits or alphanumeric\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Compliant\", \"title\": \"Passcode strength is compliant\", \"finding\": \"Passcode strength is set to alphanumeric, in line with the required >= 6 digits or alphanumeric. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Compliant\", \"title\": \"Human oversight for high-risk AI is compliant\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamonacc running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Compliant\", \"title\": \"On-access malware protection: meets the baseline\", \"finding\": \"On-access malware protection is set to clamonacc running, in line with the required on-access scanning.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (sali)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"Found: Administrator (sali). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-6F4VUQI\\jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts: does not meet the baseline\", \"finding\": \"The device reports DESKTOP-6F4VUQI\\\\jdoe, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Microsoft Copilot\nREQUIRED: inventory taken\nDETAIL: AI apps listed by the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (android):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"AI software inventory is set to Microsoft Copilot, in line with the required inventory taken. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.6 (ios)\nHOSTNAME: SALES-iPhone-15\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=6, Non-Compliant=13, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 0 detected by the browser, 21 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Non-Compliant | no passcode\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 5 minutes\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Non-Compliant | jailbroken\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Non-Compliant | off (no passcode)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.6 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 42 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.6\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Grammarly\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Grammarly\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Grammarly\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.app_sources: High\n- encryption.enabled: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=6, assessed=19, percent=32\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not jailbroken | High | jailbroken\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- encryption.enabled | Data protection (encryption) | High | off (no passcode)\n- vuln.pending_updates | iOS security updates | High | iOS 26.6 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 42 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- encryption.enabled: Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set."}, {"role": "assistant", "content": "{\"device\": \"SALES-iPhone-15 - iOS 26.6\", \"score\": {\"compliant\": 6, \"assessed\": 19, \"percent\": 32}, \"counts\": {\"Compliant\": 6, \"Non-Compliant\": 13, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not jailbroken\", \"risk\": \"High\", \"found\": \"jailbroken\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Data protection (encryption)\", \"risk\": \"High\", \"found\": \"off (no passcode)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.6 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"42 critical/high\"}], \"summary\": \"SALES-iPhone-15 meets 6 of 19 assessed checks (32%). The most serious gaps are Not jailbroken, Apps only from the App Store and Data protection (encryption). 10 items are pending. 21 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not jailbroken: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\", \"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Data protection (encryption): Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (passcode set)\nREQUIRED: on\nDETAIL: iOS encrypts the data on the phone with hardware keys once a passcode is set.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Data protection (encryption): meets the baseline\", \"finding\": \"The device reports on (passcode set), which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: LM Studio, Alpaca (Ollama client), Jan\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is LM Studio, Alpaca (Ollama client), Jan instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: rdp, chromeremotedesktop\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Found: rdp, chromeremotedesktop. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google backup\", \"finding\": \"Found: no backup. Required: backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.7 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates is non-compliant\", \"finding\": \"The device reports iOS 26.7 (update to 26.7.1 available), but the baseline requires latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: ufw active\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Compliant\", \"title\": \"Firewall is compliant\", \"finding\": \"The device reports ufw active, which meets the requirement (active).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"Found: 10 minutes. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"The device reports external drives writable without enforced encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall outbound policy\", \"finding\": \"Firewall outbound policy is no firewall instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: download only\nREQUIRED: download and install on\nDETAIL: Updates wait for the user to install them.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic iOS updates is non-compliant\", \"finding\": \"The user reports download only, but the baseline requires download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect scanning on; Gatekeeper on\nREQUIRED: XProtect and Gatekeeper on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (macos):\n- Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Compliant\", \"title\": \"Apple malware protection service matches the required setting\", \"finding\": \"Apple malware protection service is set to XProtect scanning on; Gatekeeper on, in line with the required XProtect and Gatekeeper on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Removable media encryption on macOS: Removable media encryption on macOS. Why it matters: Personal data copied to removable media must be encrypted, or copying must be blocked. Risk if missing: Medium - A lost unencrypted USB drive with personal data is a reportable breach. Requirement: Data cannot be written to unencrypted removable drives. How to fix on macOS: (1) Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives. How to verify: Re-run the audit; Removable media encryption should be Compliant. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption); GDPR Art. 34 (Communication of a personal data breach to the data subject).\n[2] ISO 27001:2022 A.7.10 Storage media: ISO 27001:2022 A.7.10 (Storage media): Manage removable media through its life cycle according to the classification of the data on it. An auditor looks for: Rules for USB and external drives; Encryption or blocking of removable media. On endpoints: Unrestricted USB writing lets data leave the device unencrypted.\n[3] GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption): Where appropriate, protect personal data with pseudonymisation and encryption. An auditor looks for: Encrypted devices and removable media. On endpoints: Disk encryption is the standard measure for laptops holding personal data.\n[4] Disk encryption protection on macOS: Disk encryption protection on macOS. Why it matters: Encryption only protects the data when it is active, fully applied and uses a strong cipher. Risk if missing: High - Suspended or partial encryption leaves data readable on a lost device. Requirement: Encryption protection is active (not suspended or in progress) with a strong cipher. How to fix on macOS: (1) Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress. How to verify: Re-run the audit; Disk encryption protection should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n\nMY AUDIT: macOS 15.6 (macos), audited 2026-10-01; 42 of 49 assessed checks compliant; 7 gaps, 3 pending, 1 not read. Gaps: USB storage write: allow; Unauthorized employee administrator accounts: sali; Known vulnerabilities in installed software: 13 critical/high; Removable media encryption: external drives writable without enforced encryption; Unapproved AI tools (shadow AI): LM Studio; Apple Intelligence external integrations restricted: not restricted; Data protection impact assessment: partial.\n\nQUESTION: How do I fix removable media encryption?"}, {"role": "assistant", "content": "Your audit found Removable media encryption: external drives writable without enforced encryption. Personal data copied to removable media must be encrypted, or copying must be blocked. To fix it [1]:\n1. Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "macos", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Microsoft Defender Antimalware [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: service True, antivirus False\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (windows):\n- Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Microsoft Defender Antimalware\", \"finding\": \"Microsoft Defender Antimalware is service True, antivirus False instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pmartin, mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"The device reports pmartin, mkhan, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Hugging Face models\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Unapproved AI tools (shadow AI) is ChatGPT, Hugging Face models instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (pmartin)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"Found: Administrator (pmartin). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"Found: no. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Human oversight for high-risk AI is no instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: protection On, XtsAes128\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"BitLocker protection: meets the baseline\", \"finding\": \"The device reports protection On, XtsAes128, which meets the requirement (protection on, AES).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"The device reports no lockout configured, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2026-05 (5 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2026-05 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Android security update\", \"finding\": \"Found: security patch 2026-05 (5 months old). Required: <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"Account lockout duration is 10 minutes instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Adobe Acrobat Reader (64-bit) Adobe.Acrobat.Reader.64-bit 5.8.533 5.1.843 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"Found: 1 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"Approved IT administrator is none of the approved accounts is an administrator instead of itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iPadOS 26.5 (ios)\nHOSTNAME: OPS-iPad-53\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=6, Non-Compliant=11, NotApplicable=0, Pending=12, Error=0\nEVIDENCE: 3 detected by the browser, 18 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Non-Compliant | no passcode\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Pending | not answered\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Non-Compliant | off (no passcode)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iPadOS 26.5 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 106 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | off\n- gdpr.os_supported | Supported operating system version | Compliant | iPadOS 26.5\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Pending | not answered\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | DeepSeek, Gemini, Replika\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | DeepSeek, Gemini, Replika\n- aiact.risk_classification | AI system risk classification | Pending | DeepSeek, Gemini, Replika\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- mobile.app_sources: High\n- encryption.enabled: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=6, assessed=17, percent=35\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- encryption.enabled | Data protection (encryption) | High | off (no passcode)\n- vuln.pending_updates | iOS security updates | High | iPadOS 26.5 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 106 critical/high\n- gdpr.auto_updates | Automatic iOS updates | High | off\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- encryption.enabled: Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered."}, {"role": "assistant", "content": "{\"device\": \"OPS-iPad-53 - iPadOS 26.5\", \"score\": {\"compliant\": 6, \"assessed\": 17, \"percent\": 35}, \"counts\": {\"Compliant\": 6, \"Non-Compliant\": 11, \"NotApplicable\": 0, \"Pending\": 12, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Data protection (encryption)\", \"risk\": \"High\", \"found\": \"off (no passcode)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iPadOS 26.5 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"106 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"off\"}], \"summary\": \"OPS-iPad-53 meets 6 of 17 assessed checks (35%). The most serious gaps are Apps only from the App Store, Data protection (encryption) and iOS security updates. 12 items are pending. 18 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Data protection (encryption): Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: USB mass storage driver disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Non-Compliant\", \"title\": \"USB storage device access is non-compliant\", \"finding\": \"Found: deny. Required: allow. Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The device access setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage device access should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 251 source packages checked against ubuntu:26.04 advisories. 33 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Compliant\", \"title\": \"Known vulnerabilities in installed software: meets the baseline\", \"finding\": \"Known vulnerabilities in installed software is set to 0 critical/high, in line with the required <= 0 critical/high.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iPadOS 18.7.9 (ios)\nAI TOOLS FOUND: Gemini, Grok\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\n- Grok: Cloud AI assistant; provider xAI (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Consumer service; needs an enterprise agreement and a transfer basis before work or personal data is entered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Grok\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer service; needs an enterprise agreement and a transfer basis before work or personal data is entered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Gemini, Grok.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nHOSTNAME: LAPTOP-V6FKZWJ\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=36, Non-Compliant=17, NotApplicable=6, Pending=5, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 20\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | 5 minutes\n- malware.antivirus | Antivirus | Compliant | Sophos Endpoint (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Non-Compliant | block\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | password-protected screen saver on for all signed-in users\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 900 seconds (machine inactivity limit)\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an enabled administrator\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (FABRIKAM\\rschmit)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 3 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 296 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 9 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26100\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: none; account management: none\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | Veeam Agent for Microsoft Windows\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | AllowTelemetry=3 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Ollama\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Ollama\n- aiact.risk_classification | AI system risk classification | Pending | Ollama\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Compliant | Recall disabled by policy\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.min_age: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- firewall.domain.inbound: High\n- firewall.private.outbound: Low\n- screen_lock.timeout: Medium\n- admin.approved: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.security_logging: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.ai_register: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=36, assessed=53, percent=68\nTOP GAPS (computed, highest risk first):\n- firewall.domain.inbound | Domain inbound | High | allow\n- vuln.pending_updates | Pending security updates | High | 2 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 296 critical/high\n- password.min_length | Minimum password length | Medium | 6\n- lockout.threshold | Account lockout threshold | Medium | 20\nFIX GUIDANCE (highest-risk gaps):\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- vuln.pending_updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-V6FKZWJ - Microsoft Windows 11 Pro 10.0.26100\", \"score\": {\"compliant\": 36, \"assessed\": 53, \"percent\": 68}, \"counts\": {\"Compliant\": 36, \"Non-Compliant\": 17, \"NotApplicable\": 6, \"Pending\": 5, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"2 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"296 critical/high\"}, {\"check_id\": \"password.min_length\", \"name\": \"Minimum password length\", \"risk\": \"Medium\", \"found\": \"6\"}, {\"check_id\": \"lockout.threshold\", \"name\": \"Account lockout threshold\", \"risk\": \"Medium\", \"found\": \"20\"}], \"summary\": \"LAPTOP-V6FKZWJ meets 36 of 53 assessed checks (68%). The most serious gaps are Domain inbound, Pending security updates and Known vulnerabilities in installed software. 5 items are pending.\", \"next_steps\": [\"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Pending security updates: Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: on\nREQUIRED: off\nDETAIL: A connected computer can read and change data on the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB debugging off\", \"finding\": \"USB debugging off is on instead of off. USB debugging lets a connected computer install apps and read data without the usual checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Anyone who can plug the unlocked phone into a computer can copy data or install software.\"}, \"recommendation\": [\"Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; USB debugging should be off.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 101 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 23 applications inventoried, 11 recognised (23 common applications covered). Google Chrome 154.0.8037.92 (66 CVEs, critical; fixed in 155.0.8059.39; e.g. CVE-2026-103622, CVE-2026-103623); Visual Studio Code 1.123.2 (19 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-50520, CVE-2026-57102); zoom.us 6.3.10 (11 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-30663, CVE-2025-30664); iTerm 3.4.18 (5 CVEs, critical; fixed in 3.4.20; e.g. CVE-2023-46300, CVE-2023-46301). 102 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"The device reports 101 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Device encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (Android 12 encrypts by default)\nREQUIRED: on\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (android):\n- Update to a supported Android version (10 or later encrypts by default) or replace the phone; set a PIN or password screen lock.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Device encryption is compliant\", \"finding\": \"Device encryption is set to on (Android 12 encrypts by default), in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access: meets the baseline\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: download and install on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Pending\", \"title\": \"Automatic iOS updates is pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 15 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period matches the required setting\", \"finding\": \"The device reports 15 minutes, which meets the requirement (>= 15 minutes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Screen timeout 10 minutes\nREQUIRED: <= 300 seconds\nDETAIL: The phone stays unlocked for 10 minutes when not in use.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen timeout: does not meet the baseline\", \"finding\": \"The user reports Screen timeout 10 minutes, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: App updates\", \"finding\": \"Found: automatic app updates off. Required: automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"Found: 12. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Real-time protection [malware.realtime] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Real-time protection scans files and processes as they are used.\nrisk: High - Without it, malware is only found by scheduled scans, after it has run.\nfix (windows):\n- Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\nverify: Re-run the audit; Real-time protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.realtime\", \"status\": \"Non-Compliant\", \"title\": \"Real-time protection is non-compliant\", \"finding\": \"Found: off. Required: on. Real-time protection scans files and processes as they are used.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without it, malware is only found by scheduled scans, after it has run.\"}, \"recommendation\": [\"Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Real-time protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.realtime", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates: does not meet the baseline\", \"finding\": \"Found: check=1 download=1 security=1 macOS-install=0. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall inbound policy\", \"finding\": \"Firewall inbound policy is firewall off, stealth mode off instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: auditd not running, journal persistent\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging: does not meet the baseline\", \"finding\": \"Security event logging is auditd not running, journal persistent instead of auditd running. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Found: ChatGPT. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.5.2 (macos)\nHOSTNAME: MKT-MacBook-23\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=22, Non-Compliant=22, NotApplicable=4, Pending=8, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Non-Compliant | not set\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5254\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Non-Compliant | XProtect 5272, 38 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode on\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 120 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | lmuller\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (jdoe)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Non-Compliant | 3 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 11 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 156 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 187 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.5.2\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Non-Compliant | Network Time: Off\n- gdpr.backup_configured | Backup configured | Compliant | ArqAgent\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | screensharing\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Cursor, Ollama, Claude\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Cursor, Ollama, Claude\n- aiact.risk_classification | AI system risk classification | Pending | Cursor, Ollama, Claude\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Intel Mac\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.signatures: High\n- firewall.enabled: High\n- firewall.inbound: High\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.time_sync: Low\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=22, assessed=44, percent=50\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Malware protection signatures | High | XProtect 5272, 38 days old\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode on\n- admin.unauthorized | Unauthorized employee administrator accounts | High | lmuller\n- user.account_type | Signed-in user account type | High | Administrator (jdoe)\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on."}, {"role": "assistant", "content": "{\"device\": \"MKT-MacBook-23 - macOS 26.5.2\", \"score\": {\"compliant\": 22, \"assessed\": 44, \"percent\": 50}, \"counts\": {\"Compliant\": 22, \"Non-Compliant\": 22, \"NotApplicable\": 4, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Malware protection signatures\", \"risk\": \"High\", \"found\": \"XProtect 5272, 38 days old\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode on\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"lmuller\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (jdoe)\"}], \"summary\": \"MKT-MacBook-23 meets 22 of 44 assessed checks (50%). The most serious gaps are Malware protection signatures, Firewall and Firewall inbound policy. 8 items are pending.\", \"next_steps\": [\"Malware protection signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\", \"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Answer the 5 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history matches the required setting\", \"finding\": \"The device reports 24, which meets the requirement (>= 24).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: root filesystem not encrypted\nREQUIRED: system disk encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (linux):\n- Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Encryption\", \"finding\": \"The device reports root filesystem not encrypted, but the baseline requires system disk encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\", \"For a stand-alone device: net accounts /maxpwage:365\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: fully encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"Disk encryption protection: meets the baseline\", \"finding\": \"Disk encryption protection is set to fully encrypted, in line with the required fully encrypted.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 26.0.1 (macos)\nAI TOOLS FOUND: Hugging Face models\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /Users/jdoe/.cache/huggingface/hub (54G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Hugging Face models.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LAPTOP-K5GG34O\\rschmit\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts: does not meet the baseline\", \"finding\": \"Found: LAPTOP-K5GG34O\\\\rschmit. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: not jailbroken (user confirmed)\nREQUIRED: not jailbroken\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Compliant\", \"title\": \"Not jailbroken: meets the baseline\", \"finding\": \"Found: not jailbroken (user confirmed). This satisfies the baseline requirement of not jailbroken. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nDETAIL: udev rule makes USB storage read-only\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Compliant\", \"title\": \"USB storage write is compliant\", \"finding\": \"The device reports deny, which meets the requirement (deny).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock is compliant\", \"finding\": \"Found: on. This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 18 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: curl, tzdata, libcurl4t64. Package lists are 3 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"The device reports 18 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Jan\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Jan.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jailbroken\nREQUIRED: not jailbroken\nDETAIL: iOS security protections are bypassed on a jailbroken phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not jailbroken: does not meet the baseline\", \"finding\": \"Found: jailbroken. Required: not jailbroken. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Gemini\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Gemini. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: AllowTelemetry=3 (Settings)\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"The device reports AllowTelemetry=3 (Settings), but the baseline requires required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store: does not meet the baseline\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Red Hat Enterprise Linux 8.10 (Ootpa)\nREQUIRED: RHEL 9 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"Found: Red Hat Enterprise Linux 8.10 (Ootpa). Required: RHEL 9 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Restrict external storage with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security intelligence [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 days old (version 1.444.986.0)\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (windows):\n- Run Update-MpSignature and check that the device can reach Windows Update or your update server.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Compliant\", \"title\": \"Security intelligence matches the required setting\", \"finding\": \"Found: 0 days old (version 1.444.986.0). This satisfies the baseline requirement of <= 3 days old.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: screen access off unless approved\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Pending\", \"title\": \"AI assistant access restricted: needs a follow-up step\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound: does not meet the baseline\", \"finding\": \"Public outbound is allow instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nDETAIL: Users checked: rschmit\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock: meets the baseline\", \"finding\": \"Found: on. This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall on, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Firewall inbound policy is firewall on, stealth mode off instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic security updates\", \"finding\": \"Automatic security updates is check=1 download=1 security=1 macOS-install=0 instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jailbroken\nREQUIRED: not jailbroken\nDETAIL: iOS security protections are bypassed on a jailbroken phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not jailbroken: does not meet the baseline\", \"finding\": \"Found: jailbroken. Required: not jailbroken. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"The device reports no, but the baseline requires yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Notification previews hidden when locked\", \"finding\": \"Notification previews hidden when locked is Always instead of When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Auto-Lock 30 seconds\nREQUIRED: <= 300 seconds\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Auto-Lock matches the required setting\", \"finding\": \"The user reports Auto-Lock 30 seconds, which meets the requirement (<= 300 seconds). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"Lockout observation period is no lockout instead of >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted is non-compliant\", \"finding\": \"Apple Intelligence external integrations restricted is not restricted instead of restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: remotemanagement\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"The device reports remotemanagement, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: within limit\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: pmartin\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout: meets the baseline\", \"finding\": \"The device reports within limit, which meets the requirement (<= 300 seconds).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 69 outdated\nREQUIRED: <= 0 outdated\nDETAIL: libpng (18.1.9) < 18.3.12; icu4c@77 (7.7.2) < 7.8.13; openssl@3 (18.9.4) < 18.10.18; xz (23.11.9) < 23.11.10; node (10.6.8) < 10.8.11.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (macos):\n- Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"The device reports 69 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: <= 7 days ago\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Error\", \"title\": \"Recent backup could not be read\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Answer ORG_ROPA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Screen lock timeout on Windows: Screen lock timeout on Windows. Why it matters: The shorter the idle time before locking, the smaller the window for misuse. Risk if missing: Medium - A long or missing timeout leaves unattended devices open. Requirement: Idle time before lock (including any grace delay) is 1..SCREEN_LOCK_MAX_SECONDS seconds (baseline: 300 = 5 minutes). Phones: Auto-Lock / screen timeout. How to fix on Windows: (1) Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less). How to verify: Re-run the audit; Screen lock timeout should be 300 seconds or less. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Screen lock password requirement on Windows: Screen lock password requirement on Windows. Why it matters: Unlocking the screen must require the user's password. Risk if missing: Medium - A screen lock that needs no password does not protect anything. Requirement: A password is required to unlock (macOS: within 5 seconds of the screen locking). How to fix on Windows: (1) Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy. How to verify: Re-run the audit; the password requirement should be Compliant. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] ISO 27001:2022 A.7.7 Clear desk and clear screen: ISO 27001:2022 A.7.7 (Clear desk and clear screen): Unattended screens and papers must not expose information. An auditor looks for: Automatic screen lock after a short idle time; A password needed to unlock. On endpoints: The screen lock settings are the technical control for clear screen.\n[4] Screen lock on Windows: Screen lock on Windows. Why it matters: An idle device should lock itself so nobody can use it while the owner is away. Risk if missing: Medium - An unlocked, unattended device exposes all the user's data and accounts. Requirement: The screen locks when idle (Windows: machine inactivity limit or a password-protected screen saver). Phones: a passcode / PIN / password screen lock is set. How to fix on Windows: (1) Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune. How to verify: Re-run the audit; Screen lock should be on. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Windows 10 or 11 (windows) on a Windows PC (ARM), Edge 141; detected by the browser, security settings not checked.\n\nQUESTION: Why does screen lock matter?"}, {"role": "assistant", "content": "An idle device should lock itself so nobody can use it while the owner is away. If it is missing, the risk is medium: An unlocked, unattended device exposes all the user's data and accounts. [4]"}], "meta": {"task": "chat", "os": "windows", "status": "why", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 8\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Minimum password length is 8 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: akaya\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"The device reports akaya, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: Apple Intelligence not available\nREQUIRED: external AI off unless approved\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Apple Intelligence external integrations restricted does not apply here\", \"finding\": \"This check does not apply: Apple Intelligence not available. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: recovery key or second unlock method\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Error\", \"title\": \"Disk encryption recovery could not be read\", \"finding\": \"No evidence was collected for this check (Requires root: run the collector with sudo.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 60 idle, password delay 300\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: 60 idle, password delay 300. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 189 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iPadOS 18.7.8 checked against NVD iOS records. iPadOS 18.7.8 (189 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2025-31255, CVE-2025-43323). 301 medium/low.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in iOS\", \"finding\": \"Known vulnerabilities in iOS is 189 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates: does not meet the baseline\", \"finding\": \"Malware signature updates is freshclam not running instead of automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 2 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-10-02  03:00:24\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Compliant\", \"title\": \"Security updates installed recently: meets the baseline\", \"finding\": \"Found: 2 days ago. This satisfies the baseline requirement of <= 30 days ago.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nHOSTNAME: LAPTOP-3TD06P0\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=30, Non-Compliant=22, NotApplicable=7, Pending=4, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | CrowdStrike Falcon (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Compliant | screen saver within limit for: CONTOSO\\pmartin\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an enabled administrator\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | LAPTOP-3TD06P0\\rschmit\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (CONTOSO\\pmartin)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 223 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | automatic updates turned off by policy\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 139 days ago\n- gdpr.os_supported | Supported operating system version | Error | not read\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: success; account management: none\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | smb\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Microsoft Copilot, Hugging Face models\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Microsoft Copilot, Hugging Face models\n- aiact.risk_classification | AI system risk classification | Pending | Microsoft Copilot, Hugging Face models\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Non-Compliant | Recall feature Enabled, not disabled by policy\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- password.complexity: Medium\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=30, assessed=52, percent=58\nTOP GAPS (computed, highest risk first):\n- firewall.public.enabled | Public firewall | High | off\n- firewall.public.inbound | Public inbound | High | allow\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | LAPTOP-3TD06P0\\rschmit\n- user.account_type | Signed-in user account type | High | Administrator (CONTOSO\\pmartin)\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 223 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- firewall.public.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.public.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- admin.unauthorized: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool)."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-3TD06P0 - Microsoft Windows 11 Enterprise 10.0.26100\", \"score\": {\"compliant\": 30, \"assessed\": 52, \"percent\": 58}, \"counts\": {\"Compliant\": 30, \"Non-Compliant\": 22, \"NotApplicable\": 7, \"Pending\": 4, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"firewall.public.enabled\", \"name\": \"Public firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.public.inbound\", \"name\": \"Public inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"LAPTOP-3TD06P0\\\\rschmit\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (CONTOSO\\\\pmartin)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"223 critical/high\"}], \"summary\": \"LAPTOP-3TD06P0 meets 30 of 52 assessed checks (58%). The most serious gaps are Public firewall, Public inbound and Unauthorized employee local administrator accounts. 1 checks could not be read. 4 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Public firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Public inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Unauthorized employee local administrator accounts: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public inbound\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically is non-compliant\", \"finding\": \"Found: set by hand. Required: set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › General › Date & Time: turn on Set Automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register is non-compliant\", \"finding\": \"Found: no. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"AI transparency to users is partial instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: none; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging: does not meet the baseline\", \"finding\": \"Found: logon: none; account management: none. Required: logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nHOSTNAME: mkt-dev-95\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=42, Non-Compliant=10, NotApplicable=2, Pending=2, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Compliant | minclass=0, required classes=4\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Non-Compliant | 80 days old\n- malware.on_access | On-access malware protection | Compliant | clamonacc running\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: pmartin (3600s)\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (pmartin)\n- encryption.enabled | Encryption | Non-Compliant | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | not encrypted\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 58 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 28 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 24.04.3 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Jan\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Jan\n- aiact.risk_classification | AI system risk classification | Pending | Jan\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- lockout.duration: Low\n- malware.signatures: High\n- screen_lock.timeout: Medium\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.outdated_apps: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- aiact.unapproved: Medium\nSCORE (computed): compliant=42, assessed=52, percent=81\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Malware signatures | High | 80 days old\n- encryption.enabled | Encryption | High | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | High | not encrypted\n- gdpr.backup_configured | Backup configured | High | none\n- screen_lock.timeout | Screen lock timeout | Medium | too long for: pmartin (3600s)\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\n- encryption.enabled: Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\n- encryption.protection: Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups."}, {"role": "assistant", "content": "{\"device\": \"mkt-dev-95 - Ubuntu 24.04.3 LTS 24.04\", \"score\": {\"compliant\": 42, \"assessed\": 52, \"percent\": 81}, \"counts\": {\"Compliant\": 42, \"Non-Compliant\": 10, \"NotApplicable\": 2, \"Pending\": 2, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"80 days old\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"root filesystem not encrypted\"}, {\"check_id\": \"encryption.protection\", \"name\": \"Disk encryption protection\", \"risk\": \"High\", \"found\": \"not encrypted\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Screen lock timeout\", \"risk\": \"Medium\", \"found\": \"too long for: pmartin (3600s)\"}], \"summary\": \"mkt-dev-95 meets 42 of 52 assessed checks (81%). The most serious gaps are Malware signatures, Encryption and Disk encryption protection. 2 items are pending.\", \"next_steps\": [\"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Encryption: Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\", \"Disk encryption protection: Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: FileVault off\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption is non-compliant\", \"finding\": \"Found: FileVault off. Required: FileVault on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 365 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration matches the required setting\", \"finding\": \"The device reports 365 days, which meets the requirement (1-365 days).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"Breach response procedure is partial instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 106 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.5 checked against NVD iOS records. iOS 26.5 (106 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-28928). 180 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS: does not meet the baseline\", \"finding\": \"Known vulnerabilities in iOS is 106 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"The device reports none found, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Microsoft Copilot, Hugging Face models\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Microsoft Copilot, Hugging Face models. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"Unauthorized employee administrator accounts is mkhan instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"Screen lock is off instead of on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAI TOOLS FOUND: Hugging Face models\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /home/jdoe/.cache/huggingface/hub (13G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Hugging Face models.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 57 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Found: 57 days ago. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: When Unlocked or Never\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Pending\", \"title\": \"Notification previews hidden when locked is pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: set automatically\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Pending\", \"title\": \"Date & time set automatically: decision pending\", \"finding\": \"The user could not find or confirm this setting. Current state: not answered. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 6-digit code\nREQUIRED: >= 6 digits or alphanumeric\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Compliant\", \"title\": \"Passcode strength matches the required setting\", \"finding\": \"Passcode strength is set to 6-digit code, in line with the required >= 6 digits or alphanumeric. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"Firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: allow\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Error\", \"title\": \"USB storage device access: could not be checked\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage device access should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no screen lock (Swipe/None)\nREQUIRED: PIN, password or pattern\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock is non-compliant\", \"finding\": \"The user reports no screen lock (Swipe/None), but the baseline requires PIN, password or pattern. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: password after 5 seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement matches the required setting\", \"finding\": \"Found: password after 5 seconds. This satisfies the baseline requirement of password within 5 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Recall disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=1; AllowRecallEnablement=0; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Compliant\", \"title\": \"Windows Recall / Copilot data analysis disabled: meets the baseline\", \"finding\": \"The device reports Recall disabled by policy, which meets the requirement (disabled by policy).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 18 applications inventoried, 12 recognised (23 common applications covered). 0 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Compliant\", \"title\": \"Known vulnerabilities in installed software is compliant\", \"finding\": \"Known vulnerabilities in installed software is set to 0 critical/high, in line with the required <= 0 critical/high.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Operating system AI features restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: none built in\nREQUIRED: restricted by policy\nDETAIL: Linux distributions ship no built-in AI assistant.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (linux):\n- No action needed: Linux distributions have no built-in AI assistant.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Operating system AI features restricted: not applicable\", \"finding\": \"Operating system AI features restricted is not assessed here: Linux distributions ship no built-in AI assistant.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain outbound: does not meet the baseline\", \"finding\": \"Domain outbound is block instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Password complexity is not enforced instead of enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Screen timeout 1 minute\nREQUIRED: <= 300 seconds\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen timeout: meets the baseline\", \"finding\": \"The user reports Screen timeout 1 minute, which meets the requirement (<= 300 seconds). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Compliant\", \"title\": \"Unauthorized employee local administrator accounts is compliant\", \"finding\": \"The device reports none, which meets the requirement (none).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; no signed-in user to check\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: no machine inactivity limit; no signed-in user to check. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Google backup is non-compliant\", \"finding\": \"Found: no backup. Required: backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 43 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Recent backup is 43 days ago instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: logon success+failure and account changes audited\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Error\", \"title\": \"Security event logging: evidence missing\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 35 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 876 source packages checked against debian:12 advisories. xymon 4.3.7-1 (11 CVEs, critical; fixed in 4.3.29-1; e.g. CVE-2015-1430, CVE-2016-2054); weborf 0.12.5-1 (1 CVE, critical; fixed in 0.19-2.1+deb12u1; e.g. CVE-2023-46586); spamassassin 3.4.4~rc1-1 (1 CVE, critical; fixed in 3.4.5~pre1-1; e.g. CVE-2020-1946); faad2 2.8.1-1 (13 CVEs, high; fixed in 2.10.0-1; e.g. CVE-2018-19502, CVE-2018-19503); modsecurity 3.0.8-1 (4 CVEs, high; fixed in 3.0.9-1+deb12u2; e.g. CVE-2023-28882, CVE-2023-38285); and 3 more. 21 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 35 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Compliant\", \"title\": \"BitLocker is compliant\", \"finding\": \"Found: on. This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history: does not meet the baseline\", \"finding\": \"Found: 5. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\", \"For a stand-alone device: net accounts /uniquepw:24\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: screen saver does not require sign-in for: FABRIKAM\\jdoe\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"Found: screen saver does not require sign-in for: FABRIKAM\\\\jdoe. Required: sign-in required to unlock. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private firewall [firewall.private.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Private firewall is non-compliant\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"Account lockout duration is no lockout instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted: does not meet the baseline\", \"finding\": \"The user reports ChatGPT extension on, but the baseline requires external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 37 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Recent backup is 37 days ago instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: ChatGPT extension off\nREQUIRED: external AI off unless approved\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Compliant\", \"title\": \"Apple Intelligence external integrations restricted matches the required setting\", \"finding\": \"Apple Intelligence external integrations restricted is set to ChatGPT extension off, in line with the required external AI off unless approved. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: Defender in Passive Mode\nREQUIRED: on\nDETAIL: Another antivirus is primary: CrowdStrike Falcon\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (windows):\n- Enable real-time protection in Microsoft Defender (on-access protection follows it).\nverify: Re-run the audit; On-access malware protection should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"NotApplicable\", \"title\": \"On-access protection: not applicable\", \"finding\": \"Not applicable on this device: Another antivirus is primary: CrowdStrike Falcon.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"Found: disabled. Required: enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 225 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-02-22 16:06:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently is non-compliant\", \"finding\": \"Security updates installed recently is 225 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public inbound\", \"finding\": \"Public inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 116 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Found: 116 days ago. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection is non-compliant\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"The device reports 10, but the baseline requires 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\", \"Set deny = 5 in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"Breach response procedure is partial instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 outdated\nREQUIRED: <= 0 outdated\nDETAIL: VLC media player VideoLAN.VLC 57.2.590 57.9.625 winget; Google Chrome Google.Chrome 34.9.131 35.7.686 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"Found: 2 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"Found: no machine inactivity limit; screen saver off. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamav-daemon\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Compliant\", \"title\": \"ClamAV malware protection service: meets the baseline\", \"finding\": \"The device reports clamav-daemon, which meets the requirement (running).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: build 22631 or later\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Error\", \"title\": \"Supported operating system version: evidence missing\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: password within 5 seconds\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Error\", \"title\": \"Screen lock password requirement: evidence missing\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store: does not meet the baseline\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ewagner\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee administrator accounts\", \"finding\": \"The device reports ewagner, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery is non-compliant\", \"finding\": \"BitLocker recovery is key protectors: Tpm instead of recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"The device reports never, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\", \"For a stand-alone device: net accounts /maxpwage:365\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The device reports Claude, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"The user reports assistant screen access on, but the baseline requires screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is ChatGPT instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: tnguyen, jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee administrator accounts\", \"finding\": \"The device reports tnguyen, jdoe, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iCloud Backup\", \"finding\": \"Found: no backup. Required: backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Sensitive notifications hidden when locked\", \"finding\": \"The user reports all content shown, but the baseline requires sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000367, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 0\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Compliant\", \"title\": \"Outdated third-party applications: meets the baseline\", \"finding\": \"The device reports 0 outdated, which meets the requirement (<= 0 outdated).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: File History (jdoe).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Found: date not readable. Confirm the last backup date in: File History (jdoe).\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: macOS 15.7.2\nREQUIRED: macOS 14 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (macos):\n- Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version is compliant\", \"finding\": \"Found: macOS 15.7.2. This satisfies the baseline requirement of macOS 14 or later.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Google Play Protect: does not meet the baseline\", \"finding\": \"Google Play Protect is Play Protect scanning off instead of on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 125 seconds\nREQUIRED: <= 300 seconds\nDETAIL: Idle 120s (screen saver/display sleep) + 5s password delay.\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout is compliant\", \"finding\": \"Screen lock timeout is set to 125 seconds, in line with the required <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: akaya, jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"The device reports akaya, jdoe, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off for: lmuller\nREQUIRED: on\nDETAIL: Users checked: lmuller\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"The device reports off for: lmuller, but the baseline requires on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 18.7 (update to 18.7.10 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 18.7.10 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates: does not meet the baseline\", \"finding\": \"The device reports iOS 18.7 (update to 18.7.10 available), but the baseline requires latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Unapproved AI tools (shadow AI) on Android: Unapproved AI tools (shadow AI) on Android. Why it matters: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Risk if missing: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations. Requirement: Every AI tool found is listed in APPROVED_AI_TOOLS. How to fix on Android: (1) Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it. (2) Publish an AI acceptable-use policy so staff know which tools are allowed. How to verify: Re-run the audit; Unapproved AI tools should read none. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 28 (Processor); GDPR Art. 44 (General principle for transfers); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[2] AI software inventory on Android: AI software inventory on Android. Why it matters: Deployers must know which AI systems they use before they can classify and govern them. Risk if missing: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools. Requirement: AI apps, local AI services and local model folders are listed (Compliant = inventory taken). How to fix on Android: (1) Add the AI tools found on this device to the organisation's AI register, with owner and purpose. How to verify: The AI register lists every tool in the inventory. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); GDPR Art. 30 (Records of processing activities); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[3] EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems): Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people. An auditor looks for: An AI register; Use according to instructions; Monitoring and log retention. On endpoints: The AI software inventory is the starting point for these duties.\n[4] AI usage log retention on Android: AI usage log retention on Android. Why it matters: Deployers of high-risk AI systems must keep the system's logs for at least six months. Risk if missing: Medium - Missing logs prevent monitoring and investigation of high-risk AI use. Requirement: Logs kept >= 6 months if a high-risk AI system is used. Pending until classification; NotApplicable when no AI tool is found. How to fix on Android: (1) If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months. How to verify: Log retention is documented for each high-risk AI system. Related controls: ISO 27001:2022 A.8.15 (Logging); EU AI Act Art. 26(6) (Deployers: keep automatically generated logs for at least six months).\n\nMY AUDIT: none\n\nMY DEVICE: Android 14 (android) on an Android phone, Chrome 141; detected by the browser, security settings not checked.\n\nQUESTION: What does EU AI Act Art. 26 require?"}, {"role": "assistant", "content": "Art. 26 (Obligations of deployers of high-risk AI systems): Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people. [3]\n\nAn auditor looks for:\n- An AI register\n- Use according to instructions\n- Monitoring and log retention\n\nOn devices: The AI software inventory is the starting point for these duties."}], "meta": {"task": "chat", "os": "any", "status": "control", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Private outbound is non-compliant\", \"finding\": \"Found: allow. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 18.7 (ios)\nAI TOOLS FOUND: ChatGPT\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Minimum password age on macOS: Minimum password age on macOS. Why it matters: A minimum age stops users cycling through passwords immediately to get back to an old one. Risk if missing: Low - Without it, the password history rule can be bypassed. Requirement: Minimum age >= PASSWORD_MIN_AGE_DAYS days. macOS has no such setting (NotApplicable). How to fix on macOS: (1) macOS has no minimum password age setting; rely on password history instead. How to verify: Re-run the audit; Minimum password age should read 1 day(s) or more. Related controls: ISO 27001:2022 A.5.17 (Authentication information); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Password history on macOS: Password history on macOS. Why it matters: Without a password history, users can return to old, possibly exposed passwords. Risk if missing: Low - Reused passwords keep compromised credentials valid. Requirement: Remembered passwords >= PASSWORD_HISTORY. How to fix on macOS: (1) Deploy a Passcode configuration profile with pinHistory = 24. How to verify: Re-run the audit; Password history should read 24 or more. Related controls: ISO 27001:2022 A.5.17 (Authentication information); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Password expiration on macOS: Password expiration on macOS. Why it matters: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful. Risk if missing: Low - A leaked password can stay valid indefinitely. Requirement: Passwords expire after 1..PASSWORD_MAX_AGE_DAYS days (0 in the baseline = expiry not required). How to fix on macOS: (1) Deploy a Passcode configuration profile with maxPINAgeInDays = 365. How to verify: Re-run the audit; Password expiration should be between 1 and 365 days. Related controls: ISO 27001:2022 A.5.17 (Authentication information); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Minimum password length on macOS: Minimum password length on macOS. Why it matters: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength. Risk if missing: Medium - Weak passwords make account takeover through guessing or credential stuffing easier. Requirement: Configured minimum length >= PASSWORD_MIN_LENGTH. How to fix on macOS: (1) Deploy a Passcode configuration profile through your MDM with minLength = 14. How to verify: Re-run the audit; Minimum password length should read 14 or more. Related controls: ISO 27001:2022 A.5.17 (Authentication information); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: macOS 14.5 (macos) on a Mac, Edge 140; detected by the browser, security settings not checked.\n\nQUESTION: What steps do I take for minimum password length on macOS?"}, {"role": "assistant", "content": "Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength. To fix it on macOS [4]:\n1. Deploy a Passcode configuration profile through your MDM with minLength = 14.\n\nTo check it worked: Re-run the audit; Minimum password length should read 14 or more."}], "meta": {"task": "chat", "os": "macos", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no passcode\nREQUIRED: passcode set\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Passcode: does not meet the baseline\", \"finding\": \"Passcode is no passcode instead of passcode set. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall outbound policy is non-compliant\", \"finding\": \"Found: no firewall. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Antivirus is non-compliant\", \"finding\": \"Antivirus is none running instead of antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength is non-compliant\", \"finding\": \"Found: 4-digit code. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 4 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-09-28 20:56:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Compliant\", \"title\": \"Security updates installed recently is compliant\", \"finding\": \"Found: 4 days ago. This satisfies the baseline requirement of <= 30 days ago.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Record of processing activities is partial instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block (firewall on, stealth mode on)\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Compliant\", \"title\": \"Firewall inbound policy matches the required setting\", \"finding\": \"The device reports block (firewall on, stealth mode on), which meets the requirement (block).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (DEVSEIS\\pmartin)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"The device reports Administrator (DEVSEIS\\\\pmartin), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: restic-backup.timer\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Compliant\", \"title\": \"Backup configured: meets the baseline\", \"finding\": \"Backup configured is set to restic-backup.timer, in line with the required backup configured.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound: does not meet the baseline\", \"finding\": \"Private inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 10 (android)\nAI TOOLS FOUND: Microsoft Copilot\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Microsoft Copilot: Cloud AI assistant; provider Microsoft (US); data: Microsoft cloud; enterprise plans keep data under Microsoft's commercial data protection terms; default risk minimal; GDPR: Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Microsoft Copilot\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Microsoft Copilot.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic system updates\", \"finding\": \"Automatic system updates is manual instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator is non-compliant\", \"finding\": \"The device reports none of the approved accounts is an administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 900 seconds (machine inactivity limit)\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"The device reports 900 seconds (machine inactivity limit), but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nHOSTNAME: dev-ubuntu-14\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=42, Non-Compliant=9, NotApplicable=2, Pending=2, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Compliant | minclass=4, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 0 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | nftables active\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | lmuller\n- user.account_type | Signed-in user account type | Compliant | Standard (akaya)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Error | not read\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 5 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 24.04.3 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Timeshift\n- gdpr.backup_recent | Recent backup | Compliant | 4 days ago\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | popularity-contest on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | LM Studio, Claude Code, Hugging Face models\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | LM Studio, Claude Code, Hugging Face models\n- aiact.risk_classification | AI system risk classification | Pending | LM Studio, Claude Code, Hugging Face models\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- lockout.observation: Low\n- malware.on_access: Medium\n- firewall.inbound: High\n- usb.write: Medium\n- admin.unauthorized: High\n- gdpr.removable_encryption: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=42, assessed=51, percent=82\nTOP GAPS (computed, highest risk first):\n- firewall.inbound | Firewall inbound policy | High | allow\n- admin.unauthorized | Unauthorized employee administrator accounts | High | lmuller\n- malware.on_access | On-access malware protection | Medium | not running\n- usb.write | USB storage write | Medium | allow\n- gdpr.removable_encryption | Removable media encryption | Medium | USB storage writable without enforced encryption\nFIX GUIDANCE (highest-risk gaps):\n- firewall.inbound: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- malware.on_access: Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning."}, {"role": "assistant", "content": "{\"device\": \"dev-ubuntu-14 - Ubuntu 24.04.3 LTS 24.04\", \"score\": {\"compliant\": 42, \"assessed\": 51, \"percent\": 82}, \"counts\": {\"Compliant\": 42, \"Non-Compliant\": 9, \"NotApplicable\": 2, \"Pending\": 2, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"lmuller\"}, {\"check_id\": \"malware.on_access\", \"name\": \"On-access malware protection\", \"risk\": \"Medium\", \"found\": \"not running\"}, {\"check_id\": \"usb.write\", \"name\": \"USB storage write\", \"risk\": \"Medium\", \"found\": \"allow\"}, {\"check_id\": \"gdpr.removable_encryption\", \"name\": \"Removable media encryption\", \"risk\": \"Medium\", \"found\": \"USB storage writable without enforced encryption\"}], \"summary\": \"dev-ubuntu-14 meets 42 of 51 assessed checks (82%). The most serious gaps are Firewall inbound policy and Unauthorized employee administrator accounts. 1 checks could not be read. 2 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Firewall inbound policy: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"On-access malware protection: Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"Found: not enforced. Required: enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: root filesystem not encrypted\nREQUIRED: system disk encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (linux):\n- Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Encryption\", \"finding\": \"Found: root filesystem not encrypted. Required: system disk encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: filesharing\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"Found: filesharing. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rooted or bootloader unlocked\nREQUIRED: not rooted\nDETAIL: Android security protections are weakened on a rooted phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not rooted: does not meet the baseline\", \"finding\": \"Not rooted is rooted or bootloader unlocked instead of not rooted. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iOS 26.7\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Safari 26.7).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version matches the required setting\", \"finding\": \"Found: iOS 26.7. This satisfies the baseline requirement of >= iOS 18.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (macos):\n- Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"The device reports none of the approved accounts is an administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-09 Cumulative Update for Windows 10 Version 22H2 for x64-based Systems (KB5043040), Security Intelligence Update for Microsoft Defender Antivirus - KB2267602 (Version 1.460.0).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Found: 2 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Password history is 5 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with pinHistory = 24.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Found: no machine inactivity limit; screen saver off. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Does the auditor need internet?: Only the first time, to download the AI auditor and the vulnerability list. After that audits and chat work offline. Pending security updates can only be checked online because they ask the vendor.\n[2] How to install on a Mac: Open the downloaded .dmg file and drag Devseis Endpoint Auditor to Applications. The first time, macOS says it could not verify the app because it is not yet notarised: click Done, then open System Settings, Privacy & Security, and click Open Anyway. After that it opens normally. It needs macOS 13 or newer.\n[3] How to run an audit of a computer: A web page cannot read a computer's settings, so computers are audited with the free Devseis desktop app for Mac, Windows (beta) and Linux. Download it from the Download page, install it, click Run audit and enter the computer password once. The first run downloads the AI auditor; the audit then takes about 5 to 15 minutes and ends with a PDF report.\n[4] Minimum password length on Linux: Minimum password length on Linux. Why it matters: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength. Risk if missing: Medium - Weak passwords make account takeover through guessing or credential stuffing easier. Requirement: Configured minimum length >= PASSWORD_MIN_LENGTH. How to fix on Linux: (1) Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack. (2) Set minlen = 14 in /etc/security/pwquality.conf. How to verify: Re-run the audit; Minimum password length should read 14 or more. Related controls: ISO 27001:2022 A.5.17 (Authentication information); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Ubuntu 22.04.5 LTS 22.04 (linux), audited 2026-10-01; 15 of 47 assessed checks compliant; 32 gaps, 7 pending, 0 not read. Gaps: Minimum password length: 8; Password history: 0; Password expiration: PASS_MAX_DAYS 99999; Minimum password age: PASS_MIN_DAYS 0; Password complexity: pam_pwquality not enabled; Account lockout threshold: 10; Account lockout duration: not set; Lockout observation period: no lockout.\n\nQUESTION: What should I fix first?"}, {"role": "assistant", "content": "Your audit found 32 gaps. Start with Minimum password length (found: 8). Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength. To fix it [4]:\n1. Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n2. Set minlen = 14 in /etc/security/pwquality.conf.\n\nThen look at: Password history, Password expiration, Minimum password age."}], "meta": {"task": "chat", "os": "linux", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked is non-compliant\", \"finding\": \"The user reports Always, but the baseline requires When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.4 (ios)\nHOSTNAME: FIN-iPhone-12\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=21, Non-Compliant=6, NotApplicable=2, Pending=0, Error=0\nEVIDENCE: 3 detected by the browser, 18 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 1 minute\n- mobile.passcode_strength | Passcode strength | Compliant | 7+ digit numeric code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store only\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.4 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 142 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.4\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | backed up to a computer\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | When Unlocked\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\n- org.ai_register: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=21, assessed=27, percent=78\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | iOS security updates | High | iOS 26.4 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 142 critical/high\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Medium | ChatGPT extension on\n- org.dpia | Data protection impact assessment | Medium | partial\n- org.ai_register | AI register | Medium | partial\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- aiact.os_ai_features: Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile."}, {"role": "assistant", "content": "{\"device\": \"FIN-iPhone-12 - iOS 26.4\", \"score\": {\"compliant\": 21, \"assessed\": 27, \"percent\": 78}, \"counts\": {\"Compliant\": 21, \"Non-Compliant\": 6, \"NotApplicable\": 2, \"Pending\": 0, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.4 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"142 critical/high\"}, {\"check_id\": \"aiact.os_ai_features\", \"name\": \"Apple Intelligence external integrations restricted\", \"risk\": \"Medium\", \"found\": \"ChatGPT extension on\"}, {\"check_id\": \"org.dpia\", \"name\": \"Data protection impact assessment\", \"risk\": \"Medium\", \"found\": \"partial\"}, {\"check_id\": \"org.ai_register\", \"name\": \"AI register\", \"risk\": \"Medium\", \"found\": \"partial\"}], \"summary\": \"FIN-iPhone-12 meets 21 of 27 assessed checks (78%). The most serious gaps are iOS security updates and Known vulnerabilities in iOS. 18 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Apple Intelligence external integrations restricted: Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"The device reports never, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\", \"For a stand-alone device: net accounts /maxpwage:365\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: backup configured\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Error\", \"title\": \"Backup configured: evidence missing\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: usb-storage module disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or use USBGuard.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Non-Compliant\", \"title\": \"USB storage read is non-compliant\", \"finding\": \"Found: deny. Required: allow. The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or use USBGuard.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage read should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 38 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 16 applications inventoried, 12 recognised (31 common applications covered). Microsoft Visual Studio Code (User) 1.109.2 (25 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-40376, CVE-2026-41109); VLC media player 2.2.6 (13 CVEs, critical; fixed in 3.0.20; e.g. CVE-2017-10699, CVE-2017-17670). 14 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"Found: 38 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (rschmit)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"The device reports Administrator (rschmit), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: screen saver does not require sign-in for: DEVSEIS\\lmuller\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"The device reports screen saver does not require sign-in for: DEVSEIS\\\\lmuller, but the baseline requires sign-in required to unlock. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Hugging Face models\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is Hugging Face models instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: filesharing, anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"The device reports filesharing, anydesk, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Domain outbound\", \"finding\": \"Found: allow. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: fully encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"Disk encryption protection is compliant\", \"finding\": \"Disk encryption protection is set to fully encrypted, in line with the required fully encrypted.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: >= 6 months for high-risk AI\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"NotApplicable\", \"title\": \"AI usage log retention does not apply here\", \"finding\": \"Not applicable on this device: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 120 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 21 applications inventoried, 12 recognised (23 common applications covered). Safari 18.6 (28 CVEs, critical, actively exploited; fixed in 26.6.1; e.g. CVE-2025-14174, CVE-2025-43342); Adobe Acrobat Reader 25.001.20529 (49 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-43550, CVE-2025-43573); Visual Studio Code 1.99.1 (30 CVEs, critical; fixed in 1.136.2; e.g. CVE-2025-21264, CVE-2025-55319); VLC 2.2.6 (13 CVEs, critical; fixed in 3.0.20; e.g. CVE-2017-10699, CVE-2017-17670). 135 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 120 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rschmit\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: rschmit. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answer ORG_AI_TRANSPARENCY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no recovery key\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery: does not meet the baseline\", \"finding\": \"The device reports no recovery key, but the baseline requires recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Zoom Workplace Zoom.Zoom 25.0.541 25.5.355 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"The device reports 1 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: >= 6 months for high-risk AI\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"NotApplicable\", \"title\": \"AI usage log retention does not apply here\", \"finding\": \"This check does not apply: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates is non-compliant\", \"finding\": \"Found: check=1 download=1 security=1 macOS-install=0. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ConfigDataInstall=0, CriticalUpdateInstall=1\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware security updates is non-compliant\", \"finding\": \"Found: ConfigDataInstall=0, CriticalUpdateInstall=1. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock is compliant\", \"finding\": \"Screen lock is set to on, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Sensitive notifications hidden when locked is non-compliant\", \"finding\": \"The user reports all content shown, but the baseline requires sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Standard (sali)\nREQUIRED: Standard\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Compliant\", \"title\": \"Signed-in user account type is compliant\", \"finding\": \"The device reports Standard (sali), which meets the requirement (Standard).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: minclass=0, required classes=0\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"The device reports minclass=0, required classes=0, but the baseline requires 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an enabled administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"Found: none of the approved accounts is an enabled administrator. Required: itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 5\nREQUIRED: 1-5\nDETAIL: pam_faillock\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Compliant\", \"title\": \"Account lockout threshold: meets the baseline\", \"finding\": \"Found: 5. This satisfies the baseline requirement of 1-5.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off for: lmuller, asmith\nREQUIRED: on\nDETAIL: Users checked: lmuller, asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"Found: off for: lmuller, asmith. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 49 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 344 packages checked against rhel:8 advisories. samba-winbind-krb5-locator-debuginfo 0:4.10.4-1.el8 (20 CVEs, critical, actively exploited; fixed in 0:4.19.4-16.el8_10; e.g. CVE-2016-2124, CVE-2020-25717); dotnet3.1 0:3.1.115-1.el8_4 (13 CVEs, critical; fixed in 0:3.1.424-1.el8_6; e.g. CVE-2020-8927, CVE-2022-24464); fence-agents-common 0:4.2.1-129.el8 (11 CVEs, high; fixed in 0:4.2.1-129.el8_10.29; e.g. CVE-2024-37891, CVE-2024-6345); python2-PyMySQL 0:0.8.0-10.module+el8.1.0+3111+de3f2d8e (5 CVEs, high; fixed in 0:0.8.0-10.module+el8.9.0+19487+7dc18407; e.g. CVE-2022-40897, CVE-2022-48560). 27 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 49 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nDETAIL: removable disk write denied by policy\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Compliant\", \"title\": \"USB storage write is compliant\", \"finding\": \"The device reports deny, which meets the requirement (deny).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nHOSTNAME: DESKTOP-IFBDINJ\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=42, Non-Compliant=17, NotApplicable=3, Pending=2, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Non-Compliant | 5 minutes\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Compliant | on\n- malware.on_access | On-access protection | Non-Compliant | off\n- malware.tamper | Tamper protection | Non-Compliant | off\n- malware.signatures | Security intelligence | Compliant | 2 days old (version 1.430.673.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; no signed-in user to check\n- screen_lock.password | Screen lock password requirement | Non-Compliant | screen saver does not require sign-in for: FABRIKAM\\jdoe\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | FABRIKAM\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (FABRIKAM\\jdoe)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 4 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 94 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 22 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 11 Pro build 22621\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | OneDrive folder backup (policy)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=1 (policy)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Non-Compliant | Recall feature Disabled, not disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.min_age: Low\n- lockout.observation: Low\n- malware.on_access: Medium\n- malware.tamper: Medium\n- firewall.private.inbound: High\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- user.account_type: High\n- encryption.protection: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- gdpr.removable_encryption: Medium\n- aiact.os_ai_features: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=42, assessed=59, percent=71\nTOP GAPS (computed, highest risk first):\n- firewall.private.inbound | Private inbound | High | allow\n- user.account_type | Signed-in user account type | High | Administrator (FABRIKAM\\jdoe)\n- encryption.protection | BitLocker protection | High | protection Off, XtsAes128\n- vuln.pending_updates | Pending security updates | High | 1 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 94 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- firewall.private.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- user.account_type: Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\n- encryption.protection: Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-IFBDINJ - Microsoft Windows 11 Pro 10.0.22621\", \"score\": {\"compliant\": 42, \"assessed\": 59, \"percent\": 71}, \"counts\": {\"Compliant\": 42, \"Non-Compliant\": 17, \"NotApplicable\": 3, \"Pending\": 2, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (FABRIKAM\\\\jdoe)\"}, {\"check_id\": \"encryption.protection\", \"name\": \"BitLocker protection\", \"risk\": \"High\", \"found\": \"protection Off, XtsAes128\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"1 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"94 critical/high\"}], \"summary\": \"DESKTOP-IFBDINJ meets 42 of 59 assessed checks (71%). The most serious gaps are Private inbound, Signed-in user account type and BitLocker protection. 2 items are pending.\", \"next_steps\": [\"Private inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Signed-in user account type: Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\", \"BitLocker protection: Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 12\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Chrome 142). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"Supported operating system version is Android 12 instead of >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"Backup configured is none found instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: enabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Guest account disabled: does not meet the baseline\", \"finding\": \"The device reports enabled, but the baseline requires disabled. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"Human oversight for high-risk AI is no instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock: meets the baseline\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: On-access malware protection\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iOS 27.0\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Safari 27.0).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version matches the required setting\", \"finding\": \"Found: iOS 27.0. This satisfies the baseline requirement of >= iOS 18.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: download and install on\nDETAIL: Security updates are not installed automatically.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic iOS updates is non-compliant\", \"finding\": \"Found: off. Required: download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google Play Protect\", \"finding\": \"Google Play Protect is Play Protect scanning off instead of on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nDETAIL: Users checked: ewagner\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock: meets the baseline\", \"finding\": \"Found: on. This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"Found: whoopsie error reporting on. Required: usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: external AI off unless approved\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Pending\", \"title\": \"Apple Intelligence external integrations restricted is pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"Found: partial. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates: does not meet the baseline\", \"finding\": \"Malware signature updates is freshclam not running instead of automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"AI register is partial instead of yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 371 packages checked against rhel:8 advisories. rsyslog-snmp 0:8.1911.0-3.el8 (1 CVE, high; fixed in 0:8.2102.0-7.el8_6.1; e.g. CVE-2022-24903). 11 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"The device reports 1 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock strength is non-compliant\", \"finding\": \"Found: 4-digit PIN. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Analytics sharing minimised\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI register\", \"finding\": \"The device reports no, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apple Intelligence external integrations restricted\", \"finding\": \"Found: ChatGPT extension on. Required: external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Claude\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: ChatGPT, Claude. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: unattended-upgrades installed=1, enabled=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates is non-compliant\", \"finding\": \"Automatic security updates is unattended-upgrades installed=1, enabled=0 instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: automatic updates turned off by policy\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates is non-compliant\", \"finding\": \"Automatic security updates is automatic updates turned off by policy instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: On-access malware protection\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall outbound policy: does not meet the baseline\", \"finding\": \"Found: no firewall. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security intelligence [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 56 days old (version 1.434.614.0)\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (windows):\n- Run Update-MpSignature and check that the device can reach Windows Update or your update server.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Security intelligence is non-compliant\", \"finding\": \"The device reports 56 days old (version 1.434.614.0), but the baseline requires <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Run Update-MpSignature and check that the device can reach Windows Update or your update server.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-UNG1VHJ\\asmith\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee local administrator accounts\", \"finding\": \"Found: DESKTOP-UNG1VHJ\\\\asmith. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"Firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Copilot, Hugging Face models\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The device reports Microsoft Copilot, Hugging Face models, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"Diagnostic data minimised is not configured instead of required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Operating system AI features restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: none built in\nREQUIRED: restricted by policy\nDETAIL: Linux distributions ship no built-in AI assistant.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (linux):\n- No action needed: Linux distributions have no built-in AI assistant.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Operating system AI features restricted: not assessed on this device\", \"finding\": \"Not applicable on this device: Linux distributions ship no built-in AI assistant.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Found: 12. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\", \"For a stand-alone device: net accounts /uniquepw:24\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"The device reports none found, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Found: no. Required: yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Data protection impact assessment is no instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain inbound is non-compliant\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Firewall inbound policy is firewall off, stealth mode off instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: GPT4All\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Found: GPT4All. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nHOSTNAME: sales-ws-73\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=34, Non-Compliant=13, NotApplicable=2, Pending=2, Error=5\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 10\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=3, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Non-Compliant | no signature database\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Error | not read\n- firewall.inbound | Firewall inbound policy | Error | not read\n- firewall.outbound | Firewall outbound policy | Error | not read\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (pmartin)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Error | not read\n- encryption.recovery | Disk encryption recovery | Error | not read\n- vuln.pending_updates | Pending security updates | Non-Compliant | 4 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 27 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 26.04.1 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Alpaca (Ollama client)\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Alpaca (Ollama client)\n- aiact.risk_classification | AI system risk classification | Pending | Alpaca (Ollama client)\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- malware.service: High\n- malware.signatures: High\n- malware.on_access: Medium\n- usb.write: Medium\n- user.account_type: High\n- vuln.pending_updates: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=34, assessed=47, percent=72\nTOP GAPS (computed, highest risk first):\n- malware.service | ClamAV malware protection service | High | not running\n- malware.signatures | Malware signatures | High | no signature database\n- user.account_type | Signed-in user account type | High | Administrator (pmartin)\n- vuln.pending_updates | Pending security updates | High | 4 pending\n- gdpr.backup_configured | Backup configured | High | none\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\n- user.account_type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account."}, {"role": "assistant", "content": "{\"device\": \"sales-ws-73 - Ubuntu 26.04.1 LTS 26.04\", \"score\": {\"compliant\": 34, \"assessed\": 47, \"percent\": 72}, \"counts\": {\"Compliant\": 34, \"Non-Compliant\": 13, \"NotApplicable\": 2, \"Pending\": 2, \"Error\": 5}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"no signature database\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (pmartin)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"4 pending\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}], \"summary\": \"sales-ws-73 meets 34 of 47 assessed checks (72%). The most serious gaps are ClamAV malware protection service, Malware signatures and Signed-in user account type. 5 checks could not be read because the audit was not run as administrator. 2 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Signed-in user account type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: sali, mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"The device reports sali, mkhan, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nDETAIL: removable disk write denied by policy\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Compliant\", \"title\": \"USB storage write: meets the baseline\", \"finding\": \"Found: deny. This satisfies the baseline requirement of deny.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off for: jdoe\nREQUIRED: on\nDETAIL: Users checked: jdoe\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock is non-compliant\", \"finding\": \"Found: off for: jdoe. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 14\nDETAIL: pwpolicy global account policy\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (macos):\n- Deploy a Passcode configuration profile through your MDM with minLength = 14.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Minimum password length is 12 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile through your MDM with minLength = 14.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 outdated\nREQUIRED: <= 0 outdated\nDETAIL: 7-Zip 24.08 (x64) 7zip.7zip 65.5.247 65.9.738 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Found: 1 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nHOSTNAME: sales-ws-11\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=17, Non-Compliant=27, NotApplicable=5, Pending=7, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 10\n- password.history | Password history | Non-Compliant | 0\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 90; accounts: jdoe (99999)\n- password.min_age | Minimum password age | Non-Compliant | PASS_MIN_DAYS 0\n- password.complexity | Password complexity | Non-Compliant | pam_pwquality not enabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | Sophos Endpoint\n- malware.service | ClamAV malware protection service | NotApplicable | not used\n- malware.signature_updates | Malware signature updates | NotApplicable | not used\n- malware.signatures | Malware signatures | NotApplicable | not used\n- malware.on_access | On-access malware protection | Compliant | Sophos Endpoint\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: mkhan\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (mkhan)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 29 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 58 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 353 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | unattended-upgrades installed=1, enabled=0\n- gdpr.updates_recent | Security updates installed recently | Compliant | 7 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Debian GNU/Linux 12 (bookworm)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Non-Compliant | NTP service no, synchronised no\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | rdp\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Ollama\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Ollama\n- aiact.risk_classification | AI system risk classification | Pending | Ollama\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.expiration: Low\n- password.min_age: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- screen_lock.password: Medium\n- usb.write: Medium\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.ai_transparency: Low\nSCORE (computed): compliant=17, assessed=44, percent=39\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\n- user.account_type | Signed-in user account type | High | Administrator (mkhan)\n- vuln.pending_updates | Pending security updates | High | 29 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 353 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\n- firewall.inbound: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\n- user.account_type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account."}, {"role": "assistant", "content": "{\"device\": \"sales-ws-11 - Debian GNU/Linux 12 (bookworm) 12\", \"score\": {\"compliant\": 17, \"assessed\": 44, \"percent\": 39}, \"counts\": {\"Compliant\": 17, \"Non-Compliant\": 27, \"NotApplicable\": 5, \"Pending\": 7, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (mkhan)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"29 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"353 critical/high\"}], \"summary\": \"sales-ws-11 meets 17 of 44 assessed checks (39%). The most serious gaps are Firewall, Firewall inbound policy and Signed-in user account type. 7 items are pending.\", \"next_steps\": [\"Firewall: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\", \"Firewall inbound policy: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\", \"Signed-in user account type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\", \"Answer the 5 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"The device reports partial, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"Found: not set. Required: 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxFailedAttempts = 5.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: App updates\", \"finding\": \"Found: automatic app updates off. Required: automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Does the auditor need internet?: Only the first time, to download the AI auditor and the vulnerability list. After that audits and chat work offline. Pending security updates can only be checked online because they ask the vendor.\n[2] How to install on a Mac: Open the downloaded .dmg file and drag Devseis Endpoint Auditor to Applications. The first time, macOS says it could not verify the app because it is not yet notarised: click Done, then open System Settings, Privacy & Security, and click Open Anyway. After that it opens normally. It needs macOS 13 or newer.\n[3] How to run an audit of a computer: A web page cannot read a computer's settings, so computers are audited with the free Devseis desktop app for Mac, Windows (beta) and Linux. Download it from the Download page, install it, click Run audit and enter the computer password once. The first run downloads the AI auditor; the audit then takes about 5 to 15 minutes and ends with a PDF report.\n[4] Lockout observation period on Windows: Lockout observation period on Windows. Why it matters: Failed attempts must be counted over a long enough window to detect slow guessing. Risk if missing: Low - A short window lets attackers pace their guesses to avoid lockout. Requirement: Failed-attempt counter kept for >= LOCKOUT_OBSERVATION_MIN_MINUTES minutes. How to fix on Windows: (1) Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more. (2) For a stand-alone device: net accounts /lockoutwindow:15 How to verify: Re-run the audit; Lockout observation period should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Microsoft Windows 11 Pro 10.0.22631 (windows), audited 2026-10-06; 39 of 54 assessed checks compliant; 15 gaps, 4 pending, 2 not read. Gaps: Lockout observation period: 5 minutes; Security intelligence: 39 days old (version 1.450.518.0); Private firewall: off; Private inbound: allow; Private outbound: allow; Public firewall: off; Public inbound: allow; Public outbound: allow.\n\nQUESTION: What should I fix first?"}, {"role": "assistant", "content": "Your audit found 15 gaps. Start with Lockout observation period (found: 5 minutes). Failed attempts must be counted over a long enough window to detect slow guessing. To fix it [4]:\n1. Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n2. For a stand-alone device: net accounts /lockoutwindow:15\n\nThen look at: Security intelligence, Private firewall, Private inbound."}], "meta": {"task": "chat", "os": "windows", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker\", \"finding\": \"The device reports off, but the baseline requires on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamonacc running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Compliant\", \"title\": \"On-access malware protection matches the required setting\", \"finding\": \"The device reports clamonacc running, which meets the requirement (on-access scanning).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: >= 6 months for high-risk AI\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"NotApplicable\", \"title\": \"AI usage log retention: not applicable\", \"finding\": \"This check does not apply: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"The device reports external drives writable without enforced encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Install unknown apps blocked is non-compliant\", \"finding\": \"Install unknown apps blocked is unknown apps allowed for at least one app instead of all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 82 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Recent backup is 82 days ago instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: no guest login\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled matches the required setting\", \"finding\": \"Found: no guest login. This satisfies the baseline requirement of no guest login.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: LightDM guest session allowed\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Guest account disabled is non-compliant\", \"finding\": \"Found: LightDM guest session allowed. Required: no guest login. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Public inbound: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"The device reports none running, but the baseline requires antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude, Microsoft Copilot\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is Claude, Microsoft Copilot instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Microsoft Defender Antimalware [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: service True, antivirus False\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (windows):\n- Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Microsoft Defender Antimalware\", \"finding\": \"Microsoft Defender Antimalware is service True, antivirus False instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 47 days old\nREQUIRED: <= 3 days old\nDETAIL: /var/lib/clamav/daily.cld\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures is non-compliant\", \"finding\": \"Malware signatures is 47 days old instead of <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 139 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-01 Cumulative Update for Windows 11 (2026-05-15)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security updates installed recently\", \"finding\": \"Found: 139 days ago. Required: <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nHOSTNAME: LAPTOP-Z2TQAKZ\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=49, Non-Compliant=10, NotApplicable=1, Pending=3, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Compliant | on\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Compliant | on\n- malware.signatures | Security intelligence | Compliant | 3 days old (version 1.432.96.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | password-protected screen saver on for all signed-in users\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; no signed-in user to check\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | DEVSEIS\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | LAPTOP-Z2TQAKZ\\jdoe\n- user.account_type | Signed-in user account type | Compliant | Standard (DEVSEIS\\mkhan)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 1 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | AUOptions=4\n- gdpr.updates_recent | Security updates installed recently | Compliant | 9 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 10 Pro build 19045\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Error | not read\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | ssh\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=0 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Hugging Face models, Ollama, Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Hugging Face models, Ollama, Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Hugging Face models, Ollama, Claude Code\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Compliant | Recall disabled by policy\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- firewall.domain.inbound: High\n- firewall.private.inbound: High\n- screen_lock.timeout: Medium\n- admin.unauthorized: High\n- vuln.outdated_apps: Medium\n- gdpr.os_supported: High\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\nSCORE (computed): compliant=49, assessed=59, percent=83\nTOP GAPS (computed, highest risk first):\n- firewall.domain.inbound | Domain inbound | High | allow\n- firewall.private.inbound | Private inbound | High | allow\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | LAPTOP-Z2TQAKZ\\jdoe\n- gdpr.os_supported | Supported operating system version | High | Microsoft Windows 10 Pro build 19045\n- gdpr.remote_access | Remote access services restricted | High | ssh\nFIX GUIDANCE (highest-risk gaps):\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- firewall.private.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- admin.unauthorized: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool)."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-Z2TQAKZ - Microsoft Windows 10 Pro 10.0.19045\", \"score\": {\"compliant\": 49, \"assessed\": 59, \"percent\": 83}, \"counts\": {\"Compliant\": 49, \"Non-Compliant\": 10, \"NotApplicable\": 1, \"Pending\": 3, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"LAPTOP-Z2TQAKZ\\\\jdoe\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Microsoft Windows 10 Pro build 19045\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"ssh\"}], \"summary\": \"LAPTOP-Z2TQAKZ meets 49 of 59 assessed checks (83%). The most serious gaps are Domain inbound, Private inbound and Unauthorized employee local administrator accounts. 1 checks could not be read. 3 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Private inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Unauthorized employee local administrator accounts: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: USB mass storage driver disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage device access\", \"finding\": \"USB storage device access is deny instead of allow. Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The device access setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage device access should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iPadOS 18.7.8\nREQUIRED: >= iOS 18\nDETAIL: Version entered by the user from Settings › General › About.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version is compliant\", \"finding\": \"Found: iPadOS 18.7.8. This satisfies the baseline requirement of >= iOS 18. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period: meets the baseline\", \"finding\": \"The device reports 30 minutes, which meets the requirement (>= 15 minutes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: The macOS application firewall filters inbound connections only.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (macos):\n- The macOS application firewall does not filter outbound traffic; use a network filter product if outbound blocking is required.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Compliant\", \"title\": \"Firewall outbound policy matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: mkhan\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts is non-compliant\", \"finding\": \"Found: mkhan. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apple Intelligence external integrations restricted\", \"finding\": \"The user reports ChatGPT extension on, but the baseline requires external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 52 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-08-10  15:22:38\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security updates installed recently\", \"finding\": \"Found: 52 days ago. Required: <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup is non-compliant\", \"finding\": \"The user reports no backup, but the baseline requires backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 183 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 18 applications inventoried, 11 recognised (23 common applications covered). Google Chrome 153.0.8010.47 (144 CVEs, critical; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301); Visual Studio Code 1.119.1 (23 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-40376, CVE-2026-45482); VLC 2.1.6 (16 CVEs, critical; fixed in 3.0.20; e.g. CVE-2016-5108, CVE-2017-17670). 167 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 183 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 240 seconds\nREQUIRED: <= 300 seconds\nDETAIL: Idle 180s (screen saver/display sleep) + 60s password delay.\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout is compliant\", \"finding\": \"Found: 240 seconds. This satisfies the baseline requirement of <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Microsoft Copilot\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"The user reports ChatGPT, Microsoft Copilot, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 26.0.1 (macos)\nAI TOOLS FOUND: Ollama, Perplexity, Jan\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local AI service listening on port 11434. Local models: /Users/jdoe/.ollama/models (56G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\n- Perplexity: Cloud AI search assistant; provider Perplexity (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs an enterprise agreement and a transfer basis before work or personal data is entered.\n- Jan: Local model runner; provider Open source (runs locally); data: On the device unless a cloud model is connected; default risk minimal; GDPR: Check whether remote API providers are configured; if so, the cloud provider rules apply.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Perplexity\", \"kind\": \"Cloud AI search assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs an enterprise agreement and a transfer basis before work or personal data is entered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Jan\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Check whether remote API providers are configured; if so, the cloud provider rules apply.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Ollama, Perplexity, Jan.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"The device reports anydesk, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Password complexity is disabled instead of enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off for: asmith\nREQUIRED: on\nDETAIL: Users checked: asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"The device reports off for: asmith, but the baseline requires on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Where to find Date & time set automatically on Android: Date & time set automatically on Android: Is 'Set time automatically' on? Where to look: Settings › System › Date & time\n[2] Date & time set automatically on Android: Date & time set automatically on Android. Why it matters: Accurate clocks are needed to put events in the right order during an investigation. Risk if missing: Low - Wrong timestamps make incident timelines unreliable. Requirement: The clock is synchronised from a time server. How to fix on Android: (1) Settings › System › Date & time: turn on Set time automatically. How to verify: Re-run the audit; Time synchronisation should be Compliant. Related controls: ISO 27001:2022 A.8.17 (Clock synchronization); GDPR Art. 33 (Notification of a personal data breach to the supervisory authority).\n[3] AI system risk classification on Android: AI system risk classification on Android. Why it matters: Each AI system must be checked against the banned practices and the high-risk uses in Annex III. Risk if missing: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards. Requirement: Each AI tool found is classified (prohibited / high-risk / limited / minimal). Done by the auditor model: Pending in the collector output, NotApplicable when no AI tool is found. How to fix on Android: (1) Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose. How to verify: Every AI tool in the inventory has a classification in the AI register. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); EU AI Act Art. 5 (Prohibited AI practices); EU AI Act Art. 6 (Classification rules for high-risk AI systems); EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)).\n[4] AI usage log retention on Android: AI usage log retention on Android. Why it matters: Deployers of high-risk AI systems must keep the system's logs for at least six months. Risk if missing: Medium - Missing logs prevent monitoring and investigation of high-risk AI use. Requirement: Logs kept >= 6 months if a high-risk AI system is used. Pending until classification; NotApplicable when no AI tool is found. How to fix on Android: (1) If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months. How to verify: Log retention is documented for each high-risk AI system. Related controls: ISO 27001:2022 A.8.15 (Logging); EU AI Act Art. 26(6) (Deployers: keep automatically generated logs for at least six months).\n\nMY AUDIT: none\n\nMY DEVICE: Android 13 (android) on a SM-S921B, Chrome 141; detected by the browser, security settings not checked.\n\nQUESTION: Which ISO 27001, GDPR or EU AI Act rules cover date & time set automatically?"}, {"role": "assistant", "content": "Date & time set automatically relates to ISO 27001:2022 A.8.17, GDPR Art. 33 [2]. Meeting the setting is evidence for these controls; it is not a certification on its own."}], "meta": {"task": "chat", "os": "android", "status": "refs", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 15 (android)\nHOSTNAME: legal-xiaomi-07\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=9, Non-Compliant=15, NotApplicable=2, Pending=4, Error=0\nEVIDENCE: 2 detected by the browser, 20 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Non-Compliant | no screen lock (Swipe/None)\n- screen_lock.timeout | Screen timeout | Non-Compliant | Screen timeout 30 minutes\n- mobile.passcode_strength | Screen lock strength | Compliant | 6-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 15 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2026-06 (4 months old)\n- vuln.outdated_apps | App updates | Pending | not answered\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Compliant | Android 15\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | phone maker cloud backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- gdpr.auto_updates: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.processor_agreements: High\n- org.ai_literacy: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=9, assessed=24, percent=38\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- vuln.pending_updates | Android security update | High | security patch 2026-06 (4 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\n- org.ropa | Record of processing activities | High | no\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone."}, {"role": "assistant", "content": "{\"device\": \"legal-xiaomi-07 - Android 15\", \"score\": {\"compliant\": 9, \"assessed\": 24, \"percent\": 38}, \"counts\": {\"Compliant\": 9, \"Non-Compliant\": 15, \"NotApplicable\": 2, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2026-06 (4 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"no\"}], \"summary\": \"legal-xiaomi-07 meets 9 of 24 assessed checks (38%). The most serious gaps are Google Play Protect, Install unknown apps blocked and Android security update. 4 items are pending. 20 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Answer the 3 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 12\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"The device reports 12, but the baseline requires >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 15 (android)\nHOSTNAME: ops-galaxy-59\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=11, Non-Compliant=8, NotApplicable=0, Pending=11, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 15 seconds\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | pattern\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 15 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2026-02 (8 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Compliant | Android 15\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | Google backup | Pending | not answered\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | no notifications on lock screen\n- aiact.inventory | AI software inventory | Compliant | ChatGPT\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- mobile.passcode_strength: Medium\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- gdpr.time_sync: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=11, assessed=19, percent=58\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- vuln.pending_updates | Android security update | High | security patch 2026-02 (8 months old)\n- mobile.passcode_strength | Screen lock strength | Medium | pattern\n- mobile.find_my_device | Find My Device | Medium | off\n- vuln.outdated_apps | App updates | Medium | automatic app updates off\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\n- mobile.passcode_strength: Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern."}, {"role": "assistant", "content": "{\"device\": \"ops-galaxy-59 - Android 15\", \"score\": {\"compliant\": 11, \"assessed\": 19, \"percent\": 58}, \"counts\": {\"Compliant\": 11, \"Non-Compliant\": 8, \"NotApplicable\": 0, \"Pending\": 11, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2026-02 (8 months old)\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Screen lock strength\", \"risk\": \"Medium\", \"found\": \"pattern\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My Device\", \"risk\": \"Medium\", \"found\": \"off\"}, {\"check_id\": \"vuln.outdated_apps\", \"name\": \"App updates\", \"risk\": \"Medium\", \"found\": \"automatic app updates off\"}], \"summary\": \"ops-galaxy-59 meets 11 of 19 assessed checks (58%). The most serious gaps are Google Play Protect and Android security update. 11 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Screen lock strength: Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 15.6 (macos)\nAI TOOLS FOUND: ChatGPT, Hugging Face models, Perplexity\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /Users/jdoe/.cache/huggingface/hub (13G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\n- Perplexity: Cloud AI search assistant; provider Perplexity (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs an enterprise agreement and a transfer basis before work or personal data is entered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Perplexity\", \"kind\": \"Cloud AI search assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs an enterprise agreement and a transfer basis before work or personal data is entered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Hugging Face models, Perplexity.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect 5301, 18 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Compliant\", \"title\": \"Malware protection signatures matches the required setting\", \"finding\": \"Found: XProtect 5301, 18 days old. This satisfies the baseline requirement of <= 30 days old.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (jdoe)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"The device reports Administrator (jdoe), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: active, strong cipher\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Error\", \"title\": \"Disk encryption protection: evidence missing\", \"finding\": \"No evidence was collected for this check (Requires root: run the collector with sudo.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iOS 26.4\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Safari 26.4).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version: meets the baseline\", \"finding\": \"The device reports iOS 26.4, which meets the requirement (>= iOS 18).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates: does not meet the baseline\", \"finding\": \"Found: freshclam not running. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: GPT4All\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: GPT4All. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: active, strong cipher\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Error\", \"title\": \"Disk encryption protection could not be read\", \"finding\": \"No evidence was collected for this check (Requires root: run the collector with sudo.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates: does not meet the baseline\", \"finding\": \"App updates is automatic app updates off instead of automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: writing to USB storage blocked\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Compliant\", \"title\": \"Removable media encryption: meets the baseline\", \"finding\": \"The device reports writing to USB storage blocked, which meets the requirement (no unencrypted writes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI register\", \"finding\": \"Found: no. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nHOSTNAME: hr-ubuntu-63\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=32, Non-Compliant=17, NotApplicable=4, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Non-Compliant | 0\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | minclass=0, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Non-Compliant | 47 days old\n- malware.on_access | On-access malware protection | Compliant | clamonacc running\n- firewall.enabled | Firewall | Compliant | firewalld active\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Non-Compliant | off for: lmuller\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: lmuller (1800s)\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: lmuller\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | ewagner\n- user.account_type | Signed-in user account type | Compliant | Standard (lmuller)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Non-Compliant | LUKS1 aes-cbc-essiv:sha256 128-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Pending | not covered\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 2 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Ubuntu 20.04.6 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | restic-backup.timer\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | anydesk\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | popularity-contest on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.complexity: Medium\n- malware.signature_updates: Medium\n- malware.signatures: High\n- firewall.inbound: High\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- admin.unauthorized: High\n- encryption.protection: High\n- gdpr.os_supported: High\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.ropa: High\n- org.dpia: Medium\n- org.processor_agreements: High\n- org.ai_transparency: Low\nSCORE (computed): compliant=32, assessed=49, percent=65\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Malware signatures | High | 47 days old\n- firewall.inbound | Firewall inbound policy | High | allow\n- admin.unauthorized | Unauthorized employee administrator accounts | High | ewagner\n- encryption.protection | Disk encryption protection | High | LUKS1 aes-cbc-essiv:sha256 128-bit key\n- gdpr.os_supported | Supported operating system version | High | Ubuntu 20.04.6 LTS\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\n- firewall.inbound: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d."}, {"role": "assistant", "content": "{\"device\": \"hr-ubuntu-63 - Ubuntu 20.04.6 LTS 20.04\", \"score\": {\"compliant\": 32, \"assessed\": 49, \"percent\": 65}, \"counts\": {\"Compliant\": 32, \"Non-Compliant\": 17, \"NotApplicable\": 4, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"47 days old\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"ewagner\"}, {\"check_id\": \"encryption.protection\", \"name\": \"Disk encryption protection\", \"risk\": \"High\", \"found\": \"LUKS1 aes-cbc-essiv:sha256 128-bit key\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Ubuntu 20.04.6 LTS\"}], \"summary\": \"hr-ubuntu-63 meets 32 of 49 assessed checks (65%). The most serious gaps are Malware signatures, Firewall inbound policy and Unauthorized employee administrator accounts. 3 items are pending.\", \"next_steps\": [\"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Firewall inbound policy: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nHOSTNAME: LAPTOP-RA4F7UO\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=51, Non-Compliant=7, NotApplicable=4, Pending=2, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Non-Compliant | off\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Compliant | on\n- malware.signatures | Security intelligence | Compliant | 0 days old (version 1.444.986.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; too long or off for: FABRIKAM\\jdoe (600s, active=0)\n- screen_lock.password | Screen lock password requirement | Compliant | machine inactivity limit locks the session\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | FABRIKAM\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | LAPTOP-RA4F7UO\\asmith\n- user.account_type | Signed-in user account type | Compliant | Standard (FABRIKAM\\jdoe)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | AUOptions=4\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 220 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 10 Pro build 19045\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | Veeam Agent for Microsoft Windows\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=1 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Non-Compliant | Recall feature Disabled, not disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- malware.realtime: High\n- screen_lock.timeout: Medium\n- admin.unauthorized: High\n- gdpr.updates_recent: High\n- gdpr.os_supported: High\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\nSCORE (computed): compliant=51, assessed=58, percent=88\nTOP GAPS (computed, highest risk first):\n- malware.realtime | Real-time protection | High | off\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | LAPTOP-RA4F7UO\\asmith\n- gdpr.updates_recent | Security updates installed recently | High | 220 days ago\n- gdpr.os_supported | Supported operating system version | High | Microsoft Windows 10 Pro build 19045\n- screen_lock.timeout | Screen lock timeout | Medium | no machine inactivity limit; too long or off for: FABRIKAM\\jdoe (600s, active=0)\nFIX GUIDANCE (highest-risk gaps):\n- malware.realtime: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\n- admin.unauthorized: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\n- gdpr.updates_recent: Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection)."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-RA4F7UO - Microsoft Windows 10 Pro 10.0.19045\", \"score\": {\"compliant\": 51, \"assessed\": 58, \"percent\": 88}, \"counts\": {\"Compliant\": 51, \"Non-Compliant\": 7, \"NotApplicable\": 4, \"Pending\": 2, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.realtime\", \"name\": \"Real-time protection\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"LAPTOP-RA4F7UO\\\\asmith\"}, {\"check_id\": \"gdpr.updates_recent\", \"name\": \"Security updates installed recently\", \"risk\": \"High\", \"found\": \"220 days ago\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Microsoft Windows 10 Pro build 19045\"}, {\"check_id\": \"screen_lock.timeout\", \"name\": \"Screen lock timeout\", \"risk\": \"Medium\", \"found\": \"no machine inactivity limit; too long or off for: FABRIKAM\\\\jdoe (600s, active=0)\"}], \"summary\": \"LAPTOP-RA4F7UO meets 51 of 58 assessed checks (88%). The most serious gaps are Real-time protection, Unauthorized employee local administrator accounts and Security updates installed recently. 2 items are pending.\", \"next_steps\": [\"Real-time protection: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\", \"Unauthorized employee local administrator accounts: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\", \"Security updates installed recently: Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery is non-compliant\", \"finding\": \"The device reports key protectors: Tpm, but the baseline requires recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude, Meta AI, Replika\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Claude, Meta AI, Replika. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 outdated\nREQUIRED: <= 0 outdated\nDETAIL: winget found no upgrades for packages it can match.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Compliant\", \"title\": \"Outdated third-party applications matches the required setting\", \"finding\": \"The device reports 0 outdated, which meets the requirement (<= 0 outdated).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery: does not meet the baseline\", \"finding\": \"Found: key protectors: Tpm. Required: recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history matches the required setting\", \"finding\": \"The device reports 24, which meets the requirement (>= 24).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 59 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-08-08  14:21:33\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently is non-compliant\", \"finding\": \"The device reports 59 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 23 applications inventoried, 13 recognised (31 common applications covered). Notepad++ (64-bit x64) 8.9.6 (2 CVEs, high; fixed in 8.9.6.1; e.g. CVE-2026-48778, CVE-2026-48800). 2 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"The device reports 2 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 42 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.6 checked against NVD iOS records. iOS 26.6 (42 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-28935). 114 medium/low.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS: does not meet the baseline\", \"finding\": \"Known vulnerabilities in iOS is 42 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Sonoma 14.8.9-25G511.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Found: 1 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Recall feature Enabled, not disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Windows Recall / Copilot data analysis disabled is non-compliant\", \"finding\": \"The device reports Recall feature Enabled, not disabled by policy, but the baseline requires disabled by policy. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: usage & diagnostics off\nREQUIRED: usage & diagnostics off\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Compliant\", \"title\": \"Usage & diagnostics sharing minimised is compliant\", \"finding\": \"The user reports usage & diagnostics off, which meets the requirement (usage & diagnostics off). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Compliant\", \"title\": \"Domain inbound: meets the baseline\", \"finding\": \"Domain inbound is set to block, in line with the required block.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off (no passcode)\nREQUIRED: on\nDETAIL: Without a passcode, iOS data protection is not active.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection (encryption)\", \"finding\": \"Data protection (encryption) is off (no passcode) instead of on. Full-disk encryption keeps data unreadable if the device is lost or stolen. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: <= 0 pending\nDETAIL: Apple's update service could not be reached.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Error\", \"title\": \"Pending security updates could not be read\", \"finding\": \"This check returned an error: Apple's update service could not be reached. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] AI assistant access restricted on Android: AI assistant access restricted on Android. Why it matters: Built-in OS AI features can capture screen content or send text to external AI services. Risk if missing: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose. Requirement: OS-level AI features that capture or send user data are disabled by policy. NotApplicable where the OS has none. Phones: external AI integrations (ChatGPT in Apple Intelligence, assistant access to screen content) are off unless approved. How to fix on Android: (1) In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it. How to verify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[2] DeepSeek (AI tool): DeepSeek: Cloud AI assistant, provider DeepSeek (China), data location: Provider cloud in the People's Republic of China. Default EU AI Act risk class for general office use: minimal. GDPR: No EU adequacy decision for China and no business data processing terms; EU data protection authorities have restricted or investigated it. Do not use with work or personal data. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[3] EU AI Act Annex III High-risk AI systems referred to in Article 6(2): EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice. An auditor looks for: No unassessed use of AI for employment decisions, credit scoring or similar Annex III purposes. On endpoints: A general AI assistant becomes high-risk if used, for example, to screen job applicants.\n[4] Where to find AI assistant access restricted on Android: AI assistant access restricted on Android: Can the AI assistant use what is on your screen? Where to look: Gemini app › your profile picture › Settings › 'Screen context' / 'Use screen context' (other assistants have similar options)\n\nMY AUDIT: Android 15 (android), audited 2026-10-06; 12 of 20 assessed checks compliant; 8 gaps, 8 pending, 0 not read. Gaps: Screen lock: no screen lock (Swipe/None); Find My Device: off; Android security update: security patch 2026-03 (7 months old); App updates: automatic app updates off; Automatic system updates: manual; Usage & diagnostics sharing minimised: usage & diagnostics on; Sensitive notifications hidden when locked: all content shown; AI assistant access restricted: assistant screen access on.\n\nQUESTION: How do I fix ai assistant access restricted?"}, {"role": "assistant", "content": "Your audit found AI assistant access restricted: assistant screen access on. Built-in OS AI features can capture screen content or send text to external AI services. To fix it [1]:\n1. In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "android", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Google backup: does not meet the baseline\", \"finding\": \"The user reports no backup, but the baseline requires backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 15 (android)\nHOSTNAME: hr-galaxy-59\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=7, Non-Compliant=12, NotApplicable=0, Pending=11, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Non-Compliant | Screen timeout 10 minutes\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | 4-digit PIN\n- mobile.jailbreak_root | Not rooted | Non-Compliant | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Non-Compliant | on\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 15 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2025-02 (20 months old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Compliant | Android 15\n- gdpr.time_sync | Date & time set automatically | Pending | not answered\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | Claude, Meta AI, Replika\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude, Meta AI, Replika\n- aiact.risk_classification | AI system risk classification | Pending | Claude, Meta AI, Replika\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.app_sources: High\n- mobile.usb_debugging: Medium\n- vuln.pending_updates: High\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=7, assessed=19, percent=37\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- mobile.jailbreak_root | Not rooted | High | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- vuln.pending_updates | Android security update | High | security patch 2025-02 (20 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- mobile.jailbreak_root: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play."}, {"role": "assistant", "content": "{\"device\": \"hr-galaxy-59 - Android 15\", \"score\": {\"compliant\": 7, \"assessed\": 19, \"percent\": 37}, \"counts\": {\"Compliant\": 7, \"Non-Compliant\": 12, \"NotApplicable\": 0, \"Pending\": 11, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not rooted\", \"risk\": \"High\", \"found\": \"rooted or bootloader unlocked\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2025-02 (20 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}], \"summary\": \"hr-galaxy-59 meets 7 of 19 assessed checks (37%). The most serious gaps are Google Play Protect, Not rooted and Install unknown apps blocked. 11 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Not rooted: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\", \"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone: does not meet the baseline\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: XProtect only\nREQUIRED: on-access scanning\nDETAIL: XProtect scans at launch and download; macOS has no built-in on-access scanning.\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (macos):\n- Not provided by macOS itself; deploy a managed endpoint protection product if required.\nverify: Re-run the audit; On-access malware protection should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"NotApplicable\", \"title\": \"On-access malware protection does not apply here\", \"finding\": \"This check does not apply: XProtect scans at launch and download; macOS has no built-in on-access scanning.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Found: no. Required: yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: disabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled matches the required setting\", \"finding\": \"Guest account disabled is set to disabled, in line with the required disabled.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.7 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates is non-compliant\", \"finding\": \"iOS security updates is iOS 26.7 (update to 26.7.1 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: LightDM guest session allowed\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Guest account disabled is non-compliant\", \"finding\": \"Guest account disabled is LightDM guest session allowed instead of no guest login. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Google Chrome Google.Chrome 65.7.876 66.9.425 winget; Notepad++ (64-bit x64) Notepad++.Notepad++ 89.3.428 90.2.681 winget; Microsoft Visual Studio Code (User) Microsoft.VisualStudioCode 15.2.132 16.4.817 winget; Slack SlackTechnologies.Slack 5.3.967 6.8.962 winget; VLC media player VideoLAN.VLC 35.4.989 35.5.826 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"Outdated third-party applications is 6 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: OneDrive folder backup (policy).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Confirm the last backup date in: OneDrive folder backup (policy). Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Compliant\", \"title\": \"Find My iPhone is compliant\", \"finding\": \"Find My iPhone is set to on, in line with the required on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 6-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Compliant\", \"title\": \"Screen lock strength is compliant\", \"finding\": \"Screen lock strength is set to 6-digit PIN, in line with the required >= 6 digits or alphanumeric. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (jdoe)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Found: Administrator (jdoe). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Domain inbound\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted is non-compliant\", \"finding\": \"The user reports ChatGPT extension on, but the baseline requires external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no FileVault\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery: does not meet the baseline\", \"finding\": \"Found: no FileVault. Required: recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"Data processing agreements is no instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 90; accounts: jdoe (99999)\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"Found: PASS_MAX_DAYS 90; accounts: jdoe (99999). Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (macos):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not assessed on this device\", \"finding\": \"This check does not apply: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Data processing agreements is no instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Block or allow external storage with your MDM (Restrictions / device control) or an endpoint DLP product.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Public firewall: does not meet the baseline\", \"finding\": \"Public firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"Found: partial. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall outbound policy\", \"finding\": \"Firewall outbound policy is no firewall instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised: does not meet the baseline\", \"finding\": \"Found: analytics sharing on. Required: analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Account lockout threshold on macOS: Account lockout threshold on macOS. Why it matters: Locking an account after a few wrong passwords stops online guessing. Risk if missing: Medium - Unlimited attempts allow brute-force attacks against local accounts. Requirement: Lockout after 1..LOCKOUT_THRESHOLD_MAX failed attempts (0 = never locks). How to fix on macOS: (1) Deploy a Passcode configuration profile with maxFailedAttempts = 5. How to verify: Re-run the audit; Account lockout threshold should be between 1 and 5. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Lockout observation period on macOS: Lockout observation period on macOS. Why it matters: Failed attempts must be counted over a long enough window to detect slow guessing. Risk if missing: Low - A short window lets attackers pace their guesses to avoid lockout. Requirement: Failed-attempt counter kept for >= LOCKOUT_OBSERVATION_MIN_MINUTES minutes. How to fix on macOS: (1) On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile. How to verify: Re-run the audit; Lockout observation period should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Account lockout duration on Windows: Account lockout duration on Windows. Why it matters: The lockout must last long enough to slow down repeated guessing. Risk if missing: Low - A short lockout lets attackers resume guessing almost at once. Requirement: Locked for >= LOCKOUT_DURATION_MIN_MINUTES minutes, or until an administrator unlocks. How to fix on Windows: (1) Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more. (2) For a stand-alone device: net accounts /lockoutduration:15 How to verify: Re-run the audit; Account lockout duration should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Account lockout duration on macOS: Account lockout duration on macOS. Why it matters: The lockout must last long enough to slow down repeated guessing. Risk if missing: Low - A short lockout lets attackers resume guessing almost at once. Requirement: Locked for >= LOCKOUT_DURATION_MIN_MINUTES minutes, or until an administrator unlocks. How to fix on macOS: (1) Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15. How to verify: Re-run the audit; Account lockout duration should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nQUESTION: What compliance controls relate to account lockout duration?"}, {"role": "assistant", "content": "Account lockout duration relates to ISO 27001:2022 A.8.5, GDPR Art. 32(1)(b) [4]. Meeting the setting is evidence for these controls; it is not a certification on its own."}], "meta": {"task": "chat", "os": "macos", "status": "refs", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is anydesk instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device is non-compliant\", \"finding\": \"Found: off. Required: on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nHOSTNAME: LAPTOP-0OAT7LM\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=40, Non-Compliant=17, NotApplicable=1, Pending=4, Error=2\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | Error | not read\n- password.complexity | Password complexity | Error | not read\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | 5 minutes\n- malware.antivirus | Antivirus | Non-Compliant | Windows Defender (off, out of date)\n- malware.service | Microsoft Defender Antimalware | Non-Compliant | service True, antivirus False\n- malware.realtime | Real-time protection | Non-Compliant | off\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Compliant | on\n- malware.signatures | Security intelligence | Non-Compliant | 56 days old (version 1.434.614.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Compliant | screen saver within limit for: DEVSEIS\\lmuller\n- screen_lock.password | Screen lock password requirement | Non-Compliant | screen saver does not require sign-in for: DEVSEIS\\lmuller\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | DEVSEIS\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | LAPTOP-0OAT7LM\\akaya\n- user.account_type | Signed-in user account type | Compliant | Standard (DEVSEIS\\lmuller)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 2 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 181 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 0 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26100\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=0 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude Code, ChatGPT, Microsoft Copilot\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude Code, ChatGPT, Microsoft Copilot\n- aiact.risk_classification | AI system risk classification | Pending | Claude Code, ChatGPT, Microsoft Copilot\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Non-Compliant | Recall feature Disabled, not disabled by policy\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.antivirus: High\n- malware.service: High\n- malware.realtime: High\n- malware.signatures: High\n- screen_lock.password: Medium\n- admin.unauthorized: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.ai_transparency: Low\nSCORE (computed): compliant=40, assessed=57, percent=70\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | Windows Defender (off, out of date)\n- malware.service | Microsoft Defender Antimalware | High | service True, antivirus False\n- malware.realtime | Real-time protection | High | off\n- malware.signatures | Security intelligence | High | 56 days old (version 1.434.614.0)\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | LAPTOP-0OAT7LM\\akaya\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\n- malware.service: Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\n- malware.realtime: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-0OAT7LM - Microsoft Windows 11 Pro 10.0.26100\", \"score\": {\"compliant\": 40, \"assessed\": 57, \"percent\": 70}, \"counts\": {\"Compliant\": 40, \"Non-Compliant\": 17, \"NotApplicable\": 1, \"Pending\": 4, \"Error\": 2}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"Windows Defender (off, out of date)\"}, {\"check_id\": \"malware.service\", \"name\": \"Microsoft Defender Antimalware\", \"risk\": \"High\", \"found\": \"service True, antivirus False\"}, {\"check_id\": \"malware.realtime\", \"name\": \"Real-time protection\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Security intelligence\", \"risk\": \"High\", \"found\": \"56 days old (version 1.434.614.0)\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"LAPTOP-0OAT7LM\\\\akaya\"}], \"summary\": \"LAPTOP-0OAT7LM meets 40 of 57 assessed checks (70%). The most serious gaps are Antivirus, Microsoft Defender Antimalware and Real-time protection. 2 checks could not be read. 4 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Antivirus: Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\", \"Microsoft Defender Antimalware: Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\", \"Real-time protection: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: only: none\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Error\", \"title\": \"Unapproved AI tools (shadow AI) could not be read\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Data processing agreements is no instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Google backup is non-compliant\", \"finding\": \"Found: no backup. Required: backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security intelligence [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 16 days old (version 1.432.45.0)\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (windows):\n- Run Update-MpSignature and check that the device can reach Windows Update or your update server.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Security intelligence: does not meet the baseline\", \"finding\": \"Found: 16 days old (version 1.432.45.0). Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Run Update-MpSignature and check that the device can reach Windows Update or your update server.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"Found: partial. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Recall feature Disabled, not disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Windows Recall / Copilot data analysis disabled\", \"finding\": \"The device reports Recall feature Disabled, not disabled by policy, but the baseline requires disabled by policy. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Install unknown apps blocked is non-compliant\", \"finding\": \"Install unknown apps blocked is unknown apps allowed for at least one app instead of all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-10 Cumulative Update for Windows 11 Version 25H2 for x64-based Systems (KB5048227).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates: does not meet the baseline\", \"finding\": \"Pending security updates is 1 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] What is in the audit report: The report has a compliance score, the most important gaps, and for each problem: what was found, why it matters, the risk level, step-by-step fixes for that operating system, how to verify the fix, and the related ISO 27001, GDPR and EU AI Act references. It also lists AI tools found and their EU AI Act risk class.\n[2] Privacy: where do audit results go?: Nowhere. Evidence and reports stay on the device. The app downloads only two things: the AI auditor model and a signed list of known vulnerabilities. Nothing about the device is uploaded. Chat history is stored only in this browser.\n[3] Automatic security updates on Windows: Automatic security updates on Windows. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on Windows: (1) Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled. How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Tamper protection on Windows: Tamper protection on Windows. Why it matters: Tamper protection stops malware or users switching off Defender. Risk if missing: Medium - Attackers commonly disable antivirus before deploying ransomware. Requirement: Defender tamper protection is on. How to fix on Windows: (1) Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell). How to verify: Re-run the audit; Tamper protection should be on. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Microsoft Windows 11 Pro 10.0.26100 (windows), audited 2026-10-06; 53 of 58 assessed checks compliant; 5 gaps, 4 pending, 0 not read. Gaps: Tamper protection: off; Signed-in user account type: Administrator (FABRIKAM\\rschmit); Remote access services restricted: winrm; Diagnostic data minimised: AllowTelemetry=3 (Settings); Unapproved AI tools (shadow AI): Claude.\n\nQUESTION: What is the most important thing to fix?"}, {"role": "assistant", "content": "Your audit found 5 gaps. Start with Tamper protection (found: off). Tamper protection stops malware or users switching off Defender. To fix it [4]:\n1. Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\n\nThen look at: Signed-in user account type, Remote access services restricted, Diagnostic data minimised."}], "meta": {"task": "chat", "os": "windows", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: type NoSync, service Stopped/Manual\nREQUIRED: NTP or domain time\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (windows):\n- w32tm /config /syncfromflags:manual /manualpeerlist:\"pool.ntp.org\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Time synchronisation\", \"finding\": \"The device reports type NoSync, service Stopped/Manual, but the baseline requires NTP or domain time. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"w32tm /config /syncfromflags:manual /manualpeerlist:\\\"pool.ntp.org\\\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Ollama, Hugging Face models\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: Ollama, Hugging Face models. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Found: USB storage writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 80 days old\nREQUIRED: <= 3 days old\nDETAIL: /var/lib/clamav/daily.cld\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures: does not meet the baseline\", \"finding\": \"Found: 80 days old. Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: NTP service yes, synchronised yes\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation: meets the baseline\", \"finding\": \"Found: NTP service yes, synchronised yes. This satisfies the baseline requirement of time synchronised.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall: does not meet the baseline\", \"finding\": \"The device reports no active firewall, but the baseline requires active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Microsoft Visual Studio Code (User) Microsoft.VisualStudioCode 100.5.135 101.1.678 winget; 7-Zip 24.08 (x64) 7zip.7zip 22.2.973 23.0.435 winget; Notepad++ (64-bit x64) Notepad++.Notepad++ 123.1.637 124.7.294 winget; VLC media player VideoLAN.VLC 58.0.358 59.3.273 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"The device reports 4 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 4\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Compliant\", \"title\": \"Account lockout threshold is compliant\", \"finding\": \"Found: 4. This satisfies the baseline requirement of 1-5.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"The device reports none of the approved accounts is an administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answer ORG_DPIA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apple Intelligence external integrations restricted\", \"finding\": \"The user reports ChatGPT extension on, but the baseline requires external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: type NTP, service Running/Manual\nREQUIRED: NTP or domain time\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (windows):\n- w32tm /config /syncfromflags:manual /manualpeerlist:\"pool.ntp.org\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation is compliant\", \"finding\": \"Found: type NTP, service Running/Manual. This satisfies the baseline requirement of NTP or domain time.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 6-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Compliant\", \"title\": \"Screen lock strength matches the required setting\", \"finding\": \"The user reports 6-digit PIN, which meets the requirement (>= 6 digits or alphanumeric). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"The user reports assistant screen access on, but the baseline requires screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nHOSTNAME: DESKTOP-U2UJTDV\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=39, Non-Compliant=19, NotApplicable=4, Pending=2, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Compliant | on\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Compliant | on\n- malware.signatures | Security intelligence | Non-Compliant | 16 days old (version 1.432.45.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Non-Compliant | block\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | machine inactivity limit set\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 900 seconds (machine inactivity limit)\n- screen_lock.password | Screen lock password requirement | Compliant | machine inactivity limit locks the session\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | FABRIKAM\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (FABRIKAM\\sali)\n- encryption.feature | BitLocker | Non-Compliant | off\n- encryption.enabled | Encryption | Non-Compliant | EncryptionInProgress (42%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, None\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 38 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | AUOptions=4\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 235 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 22631\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: none; account management: none\n- gdpr.time_sync | Time synchronisation | Non-Compliant | type NoSync, service Stopped/Manual\n- gdpr.backup_configured | Backup configured | Compliant | File History (jdoe)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | BitLocker To Go required before writing\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | anydesk, ssh\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- malware.signatures: High\n- firewall.domain.outbound: Low\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- encryption.feature: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.time_sync: Low\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.breach_procedure: High\nSCORE (computed): compliant=39, assessed=58, percent=67\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Security intelligence | High | 16 days old (version 1.432.45.0)\n- firewall.public.enabled | Public firewall | High | off\n- firewall.public.inbound | Public inbound | High | allow\n- encryption.feature | BitLocker | High | off\n- encryption.enabled | Encryption | High | EncryptionInProgress (42%)\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: Run Update-MpSignature and check that the device can reach Windows Update or your update server.\n- firewall.public.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.public.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-U2UJTDV - Microsoft Windows 11 Pro 10.0.22631\", \"score\": {\"compliant\": 39, \"assessed\": 58, \"percent\": 67}, \"counts\": {\"Compliant\": 39, \"Non-Compliant\": 19, \"NotApplicable\": 4, \"Pending\": 2, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Security intelligence\", \"risk\": \"High\", \"found\": \"16 days old (version 1.432.45.0)\"}, {\"check_id\": \"firewall.public.enabled\", \"name\": \"Public firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.public.inbound\", \"name\": \"Public inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"encryption.feature\", \"name\": \"BitLocker\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"EncryptionInProgress (42%)\"}], \"summary\": \"DESKTOP-U2UJTDV meets 39 of 58 assessed checks (67%). The most serious gaps are Security intelligence, Public firewall and Public inbound. 2 items are pending.\", \"next_steps\": [\"Security intelligence: Run Update-MpSignature and check that the device can reach Windows Update or your update server.\", \"Public firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Public inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Found: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (tnguyen)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"The device reports Administrator (tnguyen), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised: does not meet the baseline\", \"finding\": \"Usage & diagnostics sharing minimised is usage & diagnostics on instead of usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device is non-compliant\", \"finding\": \"The user reports off, but the baseline requires on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: LUKS2 aes-xts-plain64 512-bit key\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"Disk encryption protection: meets the baseline\", \"finding\": \"The device reports LUKS2 aes-xts-plain64 512-bit key, which meets the requirement (active, strong cipher).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain outbound: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 40 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 554 packages checked against rhel:8 advisories. openchange-debuginfo 0:2.3-24.el8 (3 CVEs, critical, actively exploited; fixed in 0:2.3-27.el8; e.g. CVE-2020-14318, CVE-2020-14323); krb5-libs-debuginfo 0:1.18.2-29.el8_10 (1 CVE, critical; fixed in 0:1.18.2-30.el8_10; e.g. CVE-2024-3596); python-nose 0:1.3.7-30.module+el8.1.0+3111+de3f2d8e (23 CVEs, high; fixed in 0:1.3.7-31.module+el8.10.0+20784+edafcd43; e.g. CVE-2015-20107, CVE-2020-27619); dotnet-templates-5.0 0:5.0.206-1.el8_4 (7 CVEs, high; fixed in 0:5.0.214-1.el8_6; e.g. CVE-2020-8927, CVE-2022-24464); LibRaw-devel 0:0.19.5-3.el8 (3 CVEs, high; fixed in 0:0.19.5-6.el8_10; e.g. CVE-2026-20889, CVE-2026-21413); and 2 more. 13 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"Found: 40 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 3 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-10 Cumulative Update for Windows 11 Version 24H2 for x64-based Systems (KB5044998), 2026-10 Cumulative Update for .NET Framework 3.5 and 4.8.1 for Windows 11 Version 24H2 for x64 (KB5042090), Security Intelligence Update for Microsoft Defender Antivirus - KB2267602 (Version 1.492.0).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Pending security updates is 3 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: fully encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"Disk encryption protection matches the required setting\", \"finding\": \"The device reports fully encrypted, which meets the requirement (fully encrypted).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Found: ChatGPT. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Data protection impact assessment is partial instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 90; accounts: jdoe (99999)\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"The device reports PASS_MAX_DAYS 90; accounts: jdoe (99999), but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall\", \"finding\": \"Firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically: does not meet the baseline\", \"finding\": \"Found: set by hand. Required: set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 15.7.2 (macos)\nHOSTNAME: FIN-MacBook-29\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=20, Non-Compliant=25, NotApplicable=4, Pending=7, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 10\n- password.history | Password history | Non-Compliant | not set\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5388\n- malware.service | Apple malware protection service | Non-Compliant | XProtect scanning on; Gatekeeper assessments disabled\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5301, 18 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode on\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 1800 idle, password delay 300\n- screen_lock.enabled | Screen lock | Non-Compliant | off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after off seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | ewagner, pmartin\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (ewagner)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no recovery key\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 105 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 134 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 15.7.2\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | bzserv\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | filesharing\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | LM Studio, Ollama\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | LM Studio, Ollama\n- aiact.risk_classification | AI system risk classification | Pending | LM Studio, Ollama\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.complexity: Medium\n- lockout.duration: Low\n- malware.service: High\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- screen_lock.enabled: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\nSCORE (computed): compliant=20, assessed=45, percent=44\nTOP GAPS (computed, highest risk first):\n- malware.service | Apple malware protection service | High | XProtect scanning on; Gatekeeper assessments disabled\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode on\n- admin.unauthorized | Unauthorized employee administrator accounts | High | ewagner, pmartin\n- user.account_type | Signed-in user account type | High | Administrator (ewagner)\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on."}, {"role": "assistant", "content": "{\"device\": \"FIN-MacBook-29 - macOS 15.7.2\", \"score\": {\"compliant\": 20, \"assessed\": 45, \"percent\": 44}, \"counts\": {\"Compliant\": 20, \"Non-Compliant\": 25, \"NotApplicable\": 4, \"Pending\": 7, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"Apple malware protection service\", \"risk\": \"High\", \"found\": \"XProtect scanning on; Gatekeeper assessments disabled\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode on\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"ewagner, pmartin\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (ewagner)\"}], \"summary\": \"FIN-MacBook-29 meets 20 of 45 assessed checks (44%). The most serious gaps are Apple malware protection service, Firewall and Firewall inbound policy. 7 items are pending.\", \"next_steps\": [\"Apple malware protection service: Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\", \"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Answer the 4 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain inbound is non-compliant\", \"finding\": \"Domain inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAI TOOLS FOUND: Alpaca (Ollama client)\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Alpaca (Ollama client): Desktop client for local models; provider Open source (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Alpaca (Ollama client)\", \"kind\": \"Desktop client for local models\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Alpaca (Ollama client).\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Recall feature Disabled, not disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Windows Recall / Copilot data analysis disabled: does not meet the baseline\", \"finding\": \"Found: Recall feature Disabled, not disabled by policy. Required: disabled by policy. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"Found: none running. Required: antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"The device reports 5 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (CONTOSO\\pmartin)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Found: Administrator (CONTOSO\\\\pmartin). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised is non-compliant\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption protection\", \"finding\": \"Disk encryption protection is not encrypted instead of fully encrypted. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training matches the required setting\", \"finding\": \"AI literacy training is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Found: no. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: unknown apps blocked for all apps\nREQUIRED: all apps: Not allowed\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Compliant\", \"title\": \"Install unknown apps blocked matches the required setting\", \"finding\": \"Found: unknown apps blocked for all apps. This satisfies the baseline requirement of all apps: Not allowed. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: on\nREQUIRED: off\nDETAIL: A connected computer can read and change data on the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Non-Compliant\", \"title\": \"USB debugging off is non-compliant\", \"finding\": \"USB debugging off is on instead of off. USB debugging lets a connected computer install apps and read data without the usual checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Anyone who can plug the unlocked phone into a computer can copy data or install software.\"}, \"recommendation\": [\"Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; USB debugging should be off.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history matches the required setting\", \"finding\": \"Found: 24. This satisfies the baseline requirement of >= 24.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect 5290, 3 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Compliant\", \"title\": \"Malware protection signatures is compliant\", \"finding\": \"Found: XProtect 5290, 3 days old. This satisfies the baseline requirement of <= 30 days old.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: enabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Compliant\", \"title\": \"Password complexity is compliant\", \"finding\": \"Found: enabled. This satisfies the baseline requirement of enabled.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nHOSTNAME: DESKTOP-R2S2I1D\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=34, Non-Compliant=19, NotApplicable=9, Pending=1, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | CrowdStrike Falcon (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Error | not read\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | screen saver does not require sign-in for: DESKTOP-R2S2I1D\\mkhan\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | DESKTOP-R2S2I1D\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-R2S2I1D\\jdoe, DESKTOP-R2S2I1D\\Administrator\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (DESKTOP-R2S2I1D\\mkhan)\n- encryption.feature | BitLocker | Non-Compliant | off\n- encryption.enabled | Encryption | Non-Compliant | EncryptionInProgress (42%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, None\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 349 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 22 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26100\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: none; account management: none\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Compliant | File History (jdoe)\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=1 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.feature: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.security_logging: Medium\n- gdpr.removable_encryption: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=34, assessed=53, percent=64\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | DESKTOP-R2S2I1D\\jdoe, DESKTOP-R2S2I1D\\Administrator\n- user.account_type | Signed-in user account type | High | Administrator (DESKTOP-R2S2I1D\\mkhan)\n- encryption.feature | BitLocker | High | off\n- encryption.enabled | Encryption | High | EncryptionInProgress (42%)\n- encryption.protection | BitLocker protection | High | protection Off, None\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\n- user.account_type: Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\n- encryption.feature: Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-R2S2I1D - Microsoft Windows 11 Pro 10.0.26100\", \"score\": {\"compliant\": 34, \"assessed\": 53, \"percent\": 64}, \"counts\": {\"Compliant\": 34, \"Non-Compliant\": 19, \"NotApplicable\": 9, \"Pending\": 1, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"DESKTOP-R2S2I1D\\\\jdoe, DESKTOP-R2S2I1D\\\\Administrator\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (DESKTOP-R2S2I1D\\\\mkhan)\"}, {\"check_id\": \"encryption.feature\", \"name\": \"BitLocker\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"EncryptionInProgress (42%)\"}, {\"check_id\": \"encryption.protection\", \"name\": \"BitLocker protection\", \"risk\": \"High\", \"found\": \"protection Off, None\"}], \"summary\": \"DESKTOP-R2S2I1D meets 34 of 53 assessed checks (64%). The most serious gaps are Unauthorized employee local administrator accounts, Signed-in user account type and BitLocker. 1 checks could not be read. 1 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Unauthorized employee local administrator accounts: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\", \"Signed-in user account type: Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\", \"BitLocker: Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: ChatGPT. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no FileVault\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption recovery\", \"finding\": \"Found: no FileVault. Required: recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.5 (ios)\nHOSTNAME: LEGAL-iPhone-16\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=10, Non-Compliant=8, NotApplicable=1, Pending=10, Error=0\nEVIDENCE: 3 detected by the browser, 18 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 30 seconds\n- mobile.passcode_strength | Passcode strength | Compliant | 7+ digit numeric code\n- mobile.jailbreak_root | Not jailbroken | Non-Compliant | jailbroken\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.5 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 106 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.5\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | Gemini\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Gemini\n- aiact.risk_classification | AI system risk classification | Pending | Gemini\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.jailbreak_root: High\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\nSCORE (computed): compliant=10, assessed=18, percent=56\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not jailbroken | High | jailbroken\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- vuln.pending_updates | iOS security updates | High | iOS 26.5 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 106 critical/high\n- mobile.find_my_device | Find My iPhone | Medium | off\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered."}, {"role": "assistant", "content": "{\"device\": \"LEGAL-iPhone-16 - iOS 26.5\", \"score\": {\"compliant\": 10, \"assessed\": 18, \"percent\": 56}, \"counts\": {\"Compliant\": 10, \"Non-Compliant\": 8, \"NotApplicable\": 1, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not jailbroken\", \"risk\": \"High\", \"found\": \"jailbroken\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.5 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"106 critical/high\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My iPhone\", \"risk\": \"Medium\", \"found\": \"off\"}], \"summary\": \"LEGAL-iPhone-16 meets 10 of 18 assessed checks (56%). The most serious gaps are Not jailbroken, Apps only from the App Store and iOS security updates. 10 items are pending. 18 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not jailbroken: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\", \"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates is non-compliant\", \"finding\": \"The device reports freshclam not running, but the baseline requires automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no signature database\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures: does not meet the baseline\", \"finding\": \"Malware signatures is no signature database instead of <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: XProtect 5389, 104 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware protection signatures: does not meet the baseline\", \"finding\": \"Malware protection signatures is XProtect 5389, 104 days old instead of <= 30 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: GPT4All, Ollama, Alpaca (Ollama client)\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: GPT4All, Ollama, Alpaca (Ollama client). Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption protection is non-compliant\", \"finding\": \"Disk encryption protection is not encrypted instead of fully encrypted. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: not jailbroken\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Pending\", \"title\": \"Not jailbroken: decision pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Found: 0. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: usage & diagnostics off\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Pending\", \"title\": \"Usage & diagnostics sharing minimised is pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nDETAIL: udev rule makes USB storage read-only\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Compliant\", \"title\": \"USB storage write: meets the baseline\", \"finding\": \"Found: deny. This satisfies the baseline requirement of deny.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 108 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-06-16  13:16:49\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Security updates installed recently is 108 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: usb-storage module disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or use USBGuard.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Non-Compliant\", \"title\": \"USB storage read: does not meet the baseline\", \"finding\": \"The device reports deny, but the baseline requires allow. The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or use USBGuard.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage read should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: OPS-MacBook-24\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=29, Non-Compliant=16, NotApplicable=4, Pending=7, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Non-Compliant | 0\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Non-Compliant | not set\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout\n- malware.antivirus | Antivirus | Compliant | XProtect 5396\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5395, 17 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall on, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 240 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | sali, jdoe\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (tnguyen)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 203 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Compliant | 0 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | chromeremotedesktop\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Hugging Face models\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Hugging Face models\n- aiact.risk_classification | AI system risk classification | Pending | Hugging Face models\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Intel Mac\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.expiration: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- firewall.inbound: High\n- admin.unauthorized: High\n- user.account_type: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=29, assessed=45, percent=64\nTOP GAPS (computed, highest risk first):\n- firewall.inbound | Firewall inbound policy | High | firewall on, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | sali, jdoe\n- user.account_type | Signed-in user account type | High | Administrator (tnguyen)\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 203 critical/high\n- gdpr.auto_updates | Automatic security updates | High | check=1 download=1 security=1 macOS-install=0\nFIX GUIDANCE (highest-risk gaps):\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\n- user.account_type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account."}, {"role": "assistant", "content": "{\"device\": \"OPS-MacBook-24 - macOS 26.0.1\", \"score\": {\"compliant\": 29, \"assessed\": 45, \"percent\": 64}, \"counts\": {\"Compliant\": 29, \"Non-Compliant\": 16, \"NotApplicable\": 4, \"Pending\": 7, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall on, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"sali, jdoe\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (tnguyen)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"203 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic security updates\", \"risk\": \"High\", \"found\": \"check=1 download=1 security=1 macOS-install=0\"}], \"summary\": \"OPS-MacBook-24 meets 29 of 45 assessed checks (64%). The most serious gaps are Firewall inbound policy, Unauthorized employee administrator accounts and Signed-in user account type. 7 items are pending.\", \"next_steps\": [\"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Signed-in user account type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\", \"Answer the 5 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"The device reports partial, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 144 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 21 applications inventoried, 14 recognised (31 common applications covered). Foxit PDF Reader 12.0.2 (138 CVEs, high; fixed in 2026.1.1; e.g. CVE-2022-43649, CVE-2023-27329); Microsoft Teams 2.10.1 (6 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330). 55 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Found: 144 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: DeepSeek, Grok\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: DeepSeek, Grok. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Jan\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is Jan instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: none enabled\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Compliant\", \"title\": \"Remote access services restricted: meets the baseline\", \"finding\": \"Found: none enabled. This satisfies the baseline requirement of only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 37 outdated\nREQUIRED: <= 0 outdated\nDETAIL: ca-certificates (30.18.1) < 30.19.11; jq (12.2.9) < 12.2.10; openssl@3 (22.1.4) < 22.2.10; icu4c@77 (8.0.8) < 8.0.15; gh (6.11.8) < 6.11.16.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (macos):\n- Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Found: 37 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: macOS 26.0.1\nREQUIRED: macOS 14 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (macos):\n- Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version matches the required setting\", \"finding\": \"Found: macOS 26.0.1. This satisfies the baseline requirement of macOS 14 or later.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nHOSTNAME: sales-lt-84\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=44, Non-Compliant=6, NotApplicable=2, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Compliant | minclass=3, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Non-Compliant | 60 days old\n- malware.on_access | On-access malware protection | Compliant | clamonacc running\n- firewall.enabled | Firewall | Compliant | nftables active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (mkhan)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Compliant | 23 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | restic-backup.timer\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | anydesk\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Jan, Hugging Face models\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Jan, Hugging Face models\n- aiact.risk_classification | AI system risk classification | Pending | Jan, Hugging Face models\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- malware.signatures: High\n- encryption.recovery: Medium\n- gdpr.os_supported: High\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\nSCORE (computed): compliant=44, assessed=50, percent=88\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.signatures | Malware signatures | High | 60 days old\n- gdpr.os_supported | Supported operating system version | High | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.remote_access | Remote access services restricted | High | anydesk\n- encryption.recovery | Disk encryption recovery | Medium | 1 key slots, 0 unlock tokens\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\n- gdpr.os_supported: Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance."}, {"role": "assistant", "content": "{\"device\": \"sales-lt-84 - Red Hat Enterprise Linux 8.10 (Ootpa) 8.10\", \"score\": {\"compliant\": 44, \"assessed\": 50, \"percent\": 88}, \"counts\": {\"Compliant\": 44, \"Non-Compliant\": 6, \"NotApplicable\": 2, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"60 days old\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Red Hat Enterprise Linux 8.10 (Ootpa)\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"anydesk\"}, {\"check_id\": \"encryption.recovery\", \"name\": \"Disk encryption recovery\", \"risk\": \"Medium\", \"found\": \"1 key slots, 0 unlock tokens\"}], \"summary\": \"sales-lt-84 meets 44 of 50 assessed checks (88%). The most serious gaps are Antivirus, Malware signatures and Supported operating system version. 4 items are pending.\", \"next_steps\": [\"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Supported operating system version: Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: FileVault off\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"Encryption is FileVault off instead of FileVault on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 17.7 (update to 17.7.2 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 17.7.2 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates is non-compliant\", \"finding\": \"iOS security updates is iOS 17.7 (update to 17.7.2 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator is non-compliant\", \"finding\": \"Approved IT administrator is none of the approved accounts is an administrator instead of itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Jan, Hugging Face models\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is Jan, Hugging Face models instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.7 checked against NVD iOS records. iOS 26.7 (15 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-43688). 32 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in iOS\", \"finding\": \"Found: 15 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jailbroken\nREQUIRED: not jailbroken\nDETAIL: iOS security protections are bypassed on a jailbroken phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Not jailbroken\", \"finding\": \"Found: jailbroken. Required: not jailbroken. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall: does not meet the baseline\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"Human oversight for high-risk AI is no instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in bzserv.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Confirm the last backup date in bzserv. Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: <= 300 seconds\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Pending\", \"title\": \"Screen timeout: decision pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"Found: no lockout. Required: >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption recovery\", \"finding\": \"Disk encryption recovery is 1 key slots, 0 unlock tokens instead of recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 12\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Firefox 143). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"Found: Android 12. Required: >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"The device reports 1 key slots, 0 unlock tokens, but the baseline requires recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 160 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iPadOS 18.7.9 checked against NVD iOS records. iPadOS 18.7.9 (160 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2025-31255, CVE-2025-43323). 285 medium/low.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS is non-compliant\", \"finding\": \"Found: 160 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"The device reports not set, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Install unknown apps blocked is non-compliant\", \"finding\": \"Install unknown apps blocked is unknown apps allowed for at least one app instead of all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: password after 5 seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement is compliant\", \"finding\": \"Found: password after 5 seconds. This satisfies the baseline requirement of password within 5 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1200 idle, password delay 5\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: 1200 idle, password delay 5. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Found: partial. Required: yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"The device reports none of the approved accounts is an administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: auditd not running, journal volatile\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging is non-compliant\", \"finding\": \"Security event logging is auditd not running, journal volatile instead of auditd running. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: active\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Error\", \"title\": \"Firewall: could not be checked\", \"finding\": \"No evidence was collected for this check (Requires root: run the collector with sudo.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MIN_DAYS 0\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (linux):\n- Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"Minimum password age is PASS_MIN_DAYS 0 instead of >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: OneDrive folder backup (policy).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Current state: date not readable. Confirm the last backup date in: OneDrive folder backup (policy).\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] AI system risk classification on iPhone: AI system risk classification on iPhone. Why it matters: Each AI system must be checked against the banned practices and the high-risk uses in Annex III. Risk if missing: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards. Requirement: Each AI tool found is classified (prohibited / high-risk / limited / minimal). Done by the auditor model: Pending in the collector output, NotApplicable when no AI tool is found. How to fix on iPhone: (1) Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose. How to verify: Every AI tool in the inventory has a classification in the AI register. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); EU AI Act Art. 5 (Prohibited AI practices); EU AI Act Art. 6 (Classification rules for high-risk AI systems); EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)).\n[2] EU AI Act prohibited AI practices: EU AI Act prohibited AI practices: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation.\n[3] EU AI Act Art. 5 Prohibited AI practices: EU AI Act Art. 5 (Prohibited AI practices): Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring. An auditor looks for: No AI tools used for banned purposes. On endpoints: The purpose of each detected AI tool must be checked against the banned list.\n[4] EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: EU AI Act Art. 50 (Transparency obligations for providers and deployers of certain AI systems): People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed. An auditor looks for: Disclosure practices for chatbots and AI-generated content. On endpoints: Staff publishing AI-generated content need clear rules.\n\nMY AUDIT: none\n\nMY DEVICE: iOS 18.5 (ios) on an iPhone, Safari 18.6; detected by the browser, security settings not checked.\n\nQUESTION: What AI practices are prohibited in the EU?"}, {"role": "assistant", "content": "Under the EU AI Act [2]:\n- Emotion recognition of employees or students (except for medical or safety reasons)\n- Social scoring of people\n- Manipulative or exploitative techniques that cause significant harm\n- Untargeted scraping of facial images to build face recognition databases\n- Biometric categorisation inferring sensitive traits such as political views or sexual orientation"}], "meta": {"task": "chat", "os": "any", "status": "aiact", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Hugging Face models\nREQUIRED: inventory taken\nDETAIL: Local models: /Users/jdoe/.cache/huggingface/hub (54G).\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (macos):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"Found: Hugging Face models. This satisfies the baseline requirement of inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password age\", \"finding\": \"The device reports 0 days, but the baseline requires >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: FileVault on\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Encryption: meets the baseline\", \"finding\": \"Encryption is set to FileVault on, in line with the required FileVault on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Compliant\", \"title\": \"Breach response procedure: meets the baseline\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Hugging Face models, Perplexity\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"The device reports ChatGPT, Hugging Face models, Perplexity, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect scanning on; Gatekeeper on\nREQUIRED: XProtect and Gatekeeper on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (macos):\n- Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Compliant\", \"title\": \"Apple malware protection service matches the required setting\", \"finding\": \"Found: XProtect scanning on; Gatekeeper on. This satisfies the baseline requirement of XProtect and Gatekeeper on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 11 (android)\nHOSTNAME: ops-pixel-53\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=7, Non-Compliant=11, NotApplicable=2, Pending=10, Error=0\nEVIDENCE: 2 detected by the browser, 20 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Non-Compliant | no screen lock (Swipe/None)\n- screen_lock.timeout | Screen timeout | Non-Compliant | Screen timeout 10 minutes\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | pattern\n- mobile.jailbreak_root | Not rooted | Non-Compliant | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Non-Compliant | on\n- mobile.find_my_device | Find My Device | Pending | not answered\n- encryption.enabled | Device encryption | Compliant | on (Android 11 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2025-08 (14 months old)\n- vuln.outdated_apps | App updates | Pending | not answered\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 11\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | no notifications on lock screen\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.usb_debugging: Medium\n- vuln.pending_updates: High\n- gdpr.auto_updates: High\n- gdpr.os_supported: High\n- gdpr.time_sync: Low\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=7, assessed=18, percent=39\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- mobile.jailbreak_root | Not rooted | High | rooted or bootloader unlocked\n- vuln.pending_updates | Android security update | High | security patch 2025-08 (14 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.os_supported | Supported operating system version | High | Android 11\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- mobile.jailbreak_root: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone."}, {"role": "assistant", "content": "{\"device\": \"ops-pixel-53 - Android 11\", \"score\": {\"compliant\": 7, \"assessed\": 18, \"percent\": 39}, \"counts\": {\"Compliant\": 7, \"Non-Compliant\": 11, \"NotApplicable\": 2, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not rooted\", \"risk\": \"High\", \"found\": \"rooted or bootloader unlocked\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2025-08 (14 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 11\"}], \"summary\": \"ops-pixel-53 meets 7 of 18 assessed checks (39%). The most serious gaps are Google Play Protect, Not rooted and Android security update. 10 items are pending. 20 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Not rooted: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic security updates\", \"finding\": \"Automatic security updates is check=1 download=1 security=1 macOS-install=0 instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Known vulnerabilities in installed software on Linux: Known vulnerabilities in installed software on Linux. Why it matters: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Risk if missing: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation. Requirement: Installed software versions matched against an offline copy of the OSV/NVD vulnerability data: critical or high vulnerabilities <= MAX_CRITICAL_HIGH_VULNERABILITIES. iOS: the iOS version is matched against NVD iOS/iPadOS records. How to fix on Linux: (1) Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use. (2) Use the package manager to reach a fixed version (apt / dnf / snap / flatpak). How to verify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Automatic security updates on Linux: Automatic security updates on Linux. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on Linux: (1) Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer. How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Pending security updates on Linux: Pending security updates on Linux. Why it matters: Security updates that are offered but not installed leave known, published vulnerabilities open. Risk if missing: High - Attackers target published vulnerabilities quickly after the fix is released. Requirement: Security updates offered by the OS vendor but not yet installed <= MAX_PENDING_SECURITY_UPDATES. Phones: iOS is at the latest security release known to the vulnerability data; Android security patch level is at most MOBILE_PATCH_MAX_AGE_DAYS old. How to fix on Linux: (1) Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed. How to verify: Re-run the audit; pending security updates should be 0 or fewer. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience); GDPR Art. 32(1)(d) (Security of processing: regular testing and evaluation).\n[4] Antivirus on Linux: Antivirus on Linux. Why it matters: Anti-malware software detects and blocks known malicious files and behaviour. Risk if missing: High - Without working anti-malware, ransomware and info-stealers can run unchallenged. Requirement: An enabled, up-to-date antivirus product is present. Android: Google Play Protect scanning is on. How to fix on Linux: (1) Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon. How to verify: Re-run the audit; Antivirus should be Compliant. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer, Chrome 152; detected by the browser, security settings not checked.\n\nQUESTION: My audit says 'Known vulnerabilities in installed software' is non-compliant on my Linux. What should I do?"}, {"role": "assistant", "content": "Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. To fix it on Linux [1]:\n1. Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n2. Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\n\nTo check it worked: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}], "meta": {"task": "chat", "os": "linux", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Diagnostic data minimised is whoopsie error reporting on instead of usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"Found: partial. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Domain inbound\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (windows):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory is compliant\", \"finding\": \"The device reports no AI tools found, which meets the requirement (inventory taken).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Google backup on\nREQUIRED: backup on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Compliant\", \"title\": \"Google backup is compliant\", \"finding\": \"Found: Google backup on. This satisfies the baseline requirement of backup on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no signature database\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures: does not meet the baseline\", \"finding\": \"Found: no signature database. Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.6 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates is non-compliant\", \"finding\": \"The device reports iOS 26.6 (update to 26.7.1 available), but the baseline requires latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: OneDrive folder backup (policy).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Current state: date not readable. Confirm the last backup date in: OneDrive folder backup (policy).\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: success; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging is non-compliant\", \"finding\": \"The device reports logon: success; account management: none, but the baseline requires logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Diagnostic data minimised is not configured instead of required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically: does not meet the baseline\", \"finding\": \"Date & time set automatically is set by hand instead of set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nHOSTNAME: sales-ubuntu-13\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=24, Non-Compliant=23, NotApplicable=7, Pending=1, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Non-Compliant | 12\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=3, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Error | not read\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Compliant | 0 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | firewalld active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | NotApplicable | no GNOME desktop\n- screen_lock.timeout | Screen lock timeout | NotApplicable | no GNOME desktop\n- screen_lock.password | Screen lock password requirement | NotApplicable | no GNOME desktop\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | mkhan\n- user.account_type | Signed-in user account type | Compliant | Standard (asmith)\n- encryption.enabled | Encryption | Non-Compliant | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | not encrypted\n- vuln.pending_updates | Pending security updates | Non-Compliant | 37 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 45 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 69 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 52 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Timeshift\n- gdpr.backup_recent | Recent backup | Non-Compliant | 116 days ago\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | anydesk\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.history: Low\n- lockout.observation: Low\n- malware.signature_updates: Medium\n- malware.on_access: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.os_supported: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.ropa: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.ai_register: Medium\nSCORE (computed): compliant=24, assessed=47, percent=51\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee administrator accounts | High | mkhan\n- encryption.enabled | Encryption | High | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | High | not encrypted\n- vuln.pending_updates | Pending security updates | High | 37 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 69 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- encryption.enabled: Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\n- encryption.protection: Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups."}, {"role": "assistant", "content": "{\"device\": \"sales-ubuntu-13 - Red Hat Enterprise Linux 8.10 (Ootpa) 8.10\", \"score\": {\"compliant\": 24, \"assessed\": 47, \"percent\": 51}, \"counts\": {\"Compliant\": 24, \"Non-Compliant\": 23, \"NotApplicable\": 7, \"Pending\": 1, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"mkhan\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"root filesystem not encrypted\"}, {\"check_id\": \"encryption.protection\", \"name\": \"Disk encryption protection\", \"risk\": \"High\", \"found\": \"not encrypted\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"37 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"69 critical/high\"}], \"summary\": \"sales-ubuntu-13 meets 24 of 47 assessed checks (51%). The most serious gaps are Unauthorized employee administrator accounts, Encryption and Disk encryption protection. 1 checks could not be read. 1 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Encryption: Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\", \"Disk encryption protection: Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked is non-compliant\", \"finding\": \"The user reports Always, but the baseline requires When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Compliant\", \"title\": \"Human oversight for high-risk AI: meets the baseline\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "android", "audit": 7000367, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock strength\", \"finding\": \"The user reports 4-digit PIN, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Domain inbound is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Removable media encryption is USB drives writable without encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: LM Studio, Claude Code, Hugging Face models\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"The device reports LM Studio, Claude Code, Hugging Face models, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2026-01 (9 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2026-01 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Android security update is non-compliant\", \"finding\": \"Found: security patch 2026-01 (9 months old). Required: <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Jan, Hugging Face models, Cursor\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Jan, Hugging Face models, Cursor.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 27.0 (ios)\nHOSTNAME: FIN-iPhone-45\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=9, Non-Compliant=10, NotApplicable=2, Pending=8, Error=0\nEVIDENCE: 3 detected by the browser, 18 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Non-Compliant | no passcode\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Non-Compliant | off (no passcode)\n- vuln.pending_updates | iOS security updates | Compliant | iOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | off\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 27.0\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | When Unlocked\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- encryption.enabled: High\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=9, assessed=19, percent=47\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- encryption.enabled | Data protection (encryption) | High | off (no passcode)\n- gdpr.auto_updates | Automatic iOS updates | High | off\n- gdpr.backup_configured | iCloud Backup | High | no backup\n- screen_lock.enabled | Passcode | Medium | no passcode\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- encryption.enabled: Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\n- gdpr.auto_updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files."}, {"role": "assistant", "content": "{\"device\": \"FIN-iPhone-45 - iOS 27.0\", \"score\": {\"compliant\": 9, \"assessed\": 19, \"percent\": 47}, \"counts\": {\"Compliant\": 9, \"Non-Compliant\": 10, \"NotApplicable\": 2, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Data protection (encryption)\", \"risk\": \"High\", \"found\": \"off (no passcode)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"screen_lock.enabled\", \"name\": \"Passcode\", \"risk\": \"Medium\", \"found\": \"no passcode\"}], \"summary\": \"FIN-iPhone-45 meets 9 of 19 assessed checks (47%). The most serious gaps are Apps only from the App Store, Data protection (encryption) and Automatic iOS updates. 8 items are pending. 18 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Data protection (encryption): Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\", \"Automatic iOS updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 15 (android)\nAI TOOLS FOUND: Grok\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Grok: Cloud AI assistant; provider xAI (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Consumer service; needs an enterprise agreement and a transfer basis before work or personal data is entered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Grok\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer service; needs an enterprise agreement and a transfer basis before work or personal data is entered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Grok.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised is non-compliant\", \"finding\": \"Found: usage & diagnostics on. Required: usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after off seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"Found: password after off seconds. Required: password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 94 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 24 applications inventoried, 13 recognised (31 common applications covered). Mozilla Firefox (x64 en-US) 155.0.0 (94 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100756, CVE-2026-100767). 14 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"The device reports 94 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength: does not meet the baseline\", \"finding\": \"Passcode strength is 4-digit code instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Network Time: Off\nREQUIRED: network time on\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"Found: Network Time: Off. Required: network time on. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"sudo systemsetup -setusingnetworktime on.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: App updates\", \"finding\": \"Found: automatic app updates off. Required: automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"AI literacy training is partial instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength is non-compliant\", \"finding\": \"Passcode strength is 4-digit code instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; no signed-in user to check\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Found: no machine inactivity limit; no signed-in user to check. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 12\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Firefox 143). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"Supported operating system version is Android 12 instead of >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAI TOOLS FOUND: LM Studio, Alpaca (Ollama client), Jan\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- LM Studio: Local model runner; provider Element Labs (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences and whether its local server is exposed.\n- Alpaca (Ollama client): Desktop client for local models; provider Open source (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences.\n- Jan: Local model runner; provider Open source (runs locally); data: On the device unless a cloud model is connected; default risk minimal; GDPR: Check whether remote API providers are configured; if so, the cloud provider rules apply.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"LM Studio\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences and whether its local server is exposed.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Alpaca (Ollama client)\", \"kind\": \"Desktop client for local models\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Jan\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Check whether remote API providers are configured; if so, the cloud provider rules apply.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: LM Studio, Alpaca (Ollama client), Jan.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MIN_DAYS 0\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (linux):\n- Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age is non-compliant\", \"finding\": \"The device reports PASS_MIN_DAYS 0, but the baseline requires >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates is non-compliant\", \"finding\": \"The user reports automatic app updates off, but the baseline requires automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 69 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-07-29  13:38:41\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security updates installed recently\", \"finding\": \"Found: 69 days ago. Required: <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Tamper protection\", \"finding\": \"The device reports off, but the baseline requires on. Tamper protection stops malware or users switching off Defender.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Attackers commonly disable antivirus before deploying ransomware.\"}, \"recommendation\": [\"Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Tamper protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: auditd running, journal persistent\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging matches the required setting\", \"finding\": \"Security event logging is set to auditd running, journal persistent, in line with the required auditd running.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: automatic updates turned off by policy\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates is non-compliant\", \"finding\": \"Found: automatic updates turned off by policy. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Breach response procedure is partial instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold: does not meet the baseline\", \"finding\": \"Account lockout threshold is 0 instead of 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: sign-in required to unlock\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Error\", \"title\": \"Screen lock password requirement: evidence missing\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Compliant\", \"title\": \"Record of processing activities is compliant\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"Found: 10 minutes. Required: >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"Password expiration is never instead of 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: Apple Intelligence not available\nREQUIRED: external AI off unless approved\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Apple Intelligence external integrations restricted does not apply here\", \"finding\": \"Not applicable on this device: Apple Intelligence not available. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: on\nREQUIRED: off\nDETAIL: A connected computer can read and change data on the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Non-Compliant\", \"title\": \"USB debugging off is non-compliant\", \"finding\": \"The user reports on, but the baseline requires off. USB debugging lets a connected computer install apps and read data without the usual checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Anyone who can plug the unlocked phone into a computer can copy data or install software.\"}, \"recommendation\": [\"Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; USB debugging should be off.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Found: partial. Required: yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Found: USB storage writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold: does not meet the baseline\", \"finding\": \"Found: 0. Required: 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: Defender in Passive Mode\nREQUIRED: on\nDETAIL: Another antivirus is primary: Sophos Endpoint\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"NotApplicable\", \"title\": \"Tamper protection: not applicable\", \"finding\": \"This check does not apply: Another antivirus is primary: Sophos Endpoint.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: LightDM guest session allowed\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Guest account disabled\", \"finding\": \"Found: LightDM guest session allowed. Required: no guest login. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: ChatGPT, Claude\nREQUIRED: inventory taken\nDETAIL: AI apps listed by the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (android):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"The user reports ChatGPT, Claude, which meets the requirement (inventory taken). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold: does not meet the baseline\", \"finding\": \"The device reports 10, but the baseline requires 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\", \"Set deny = 5 in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off for: tnguyen\nREQUIRED: on\nDETAIL: Users checked: tnguyen\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"The device reports off for: tnguyen, but the baseline requires on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iPadOS 18.7.8 (update to 18.7.10 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iPadOS 18.7.10 or later in Settings › General › Software Update.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates: does not meet the baseline\", \"finding\": \"iOS security updates is iPadOS 18.7.8 (update to 18.7.10 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"AI transparency to users is no instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store is non-compliant\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, None\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker protection: does not meet the baseline\", \"finding\": \"The device reports protection Off, None, but the baseline requires protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: akaya (1800s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: akaya\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Found: too long for: akaya (1800s). Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 10 Pro build 19045\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"The device reports Microsoft Windows 10 Pro build 19045, but the baseline requires build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Found: firewall off, stealth mode off. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: ClamAV (clamav-daemon)\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus matches the required setting\", \"finding\": \"The device reports ClamAV (clamav-daemon), which meets the requirement (antivirus running).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Encryption on macOS: Encryption on macOS. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: The system disk is fully encrypted (BitLocker / LUKS / FileVault). Phones: iOS data protection is on when a passcode is set; Android 10+ encrypts by default. How to fix on macOS: (1) Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[2] ISO 27001:2022 A.5.12 Classification of information: ISO 27001:2022 A.5.12 (Classification of information): Label information by how sensitive it is so that the right protection can be applied. An auditor looks for: A classification scheme; Evidence that sensitive data is identified, e.g. personal data. On endpoints: Personal or confidential files stored on laptops need to be found and protected accordingly.\n[3] GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption): Where appropriate, protect personal data with pseudonymisation and encryption. An auditor looks for: Encrypted devices and removable media. On endpoints: Disk encryption is the standard measure for laptops holding personal data.\n[4] ISO 27001:2022 A.8.20 Networks security: ISO 27001:2022 A.8.20 (Networks security): Secure, manage and control networks and network devices to protect information. An auditor looks for: Host firewalls enabled; Unsolicited inbound connections blocked. On endpoints: The host firewall protects the laptop on untrusted networks such as hotels and cafés.\n\nMY AUDIT: none\n\nMY DEVICE: macOS (macos) on a Mac, Chrome 141; detected by the browser, security settings not checked.\n\nQUESTION: What's the best gaming laptop?"}, {"role": "assistant", "content": "I can only help with device security, ISO 27001, GDPR and the EU AI Act, so I can't answer that. Ask me, for example, how to turn on disk encryption, whether an AI tool is OK for work, or run an audit of this device."}], "meta": {"task": "chat", "os": "any", "status": "oos", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 11 Pro build 22621\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"Found: Microsoft Windows 11 Pro build 22621. Required: build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Sensitive notifications hidden when locked: does not meet the baseline\", \"finding\": \"The user reports all content shown, but the baseline requires sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"Found: no lockout. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Windows Defender (on, up to date)\nREQUIRED: enabled and up to date\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (windows):\n- Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus is compliant\", \"finding\": \"The device reports Windows Defender (on, up to date), which meets the requirement (enabled and up to date).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1800 idle, password delay 300\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Screen lock timeout is 1800 idle, password delay 300 instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (passcode set)\nREQUIRED: on\nDETAIL: iOS encrypts the data on the phone with hardware keys once a passcode is set.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Data protection (encryption) matches the required setting\", \"finding\": \"Data protection (encryption) is set to on (passcode set), in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"Account lockout duration is no lockout instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"The device reports share Mac analytics=1, share with app developers=0, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: AllowTelemetry=1 (policy)\nREQUIRED: required diagnostic data only (0-1)\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Compliant\", \"title\": \"Diagnostic data minimised matches the required setting\", \"finding\": \"Diagnostic data minimised is set to AllowTelemetry=1 (policy), in line with the required required diagnostic data only (0-1).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Answer ORG_AI_TRANSPARENCY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 1 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Compliant\", \"title\": \"Minimum password age matches the required setting\", \"finding\": \"Found: 1 days. This satisfies the baseline requirement of >= 1 days.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Malware security updates on macOS: Malware security updates on macOS. Why it matters: Signature updates must happen automatically so protection stays current without user action. Risk if missing: Medium - Manual updates are forgotten, leaving protection outdated. Requirement: Automatic signature/security-data updates are enabled. How to fix on macOS: (1) Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile. How to verify: Re-run the audit; signature updates should be automatic. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Malware protection signatures on macOS: Malware protection signatures on macOS. Why it matters: Anti-malware can only detect threats it knows about; signatures must be current. Risk if missing: High - Outdated signatures miss recent malware. Requirement: Signatures no older than SIGNATURE_MAX_AGE_DAYS (macOS XProtect: MACOS_XPROTECT_MAX_AGE_DAYS). How to fix on macOS: (1) Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now. How to verify: Re-run the audit; signatures should be within the allowed age. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] How to run an audit of a computer: A web page cannot read a computer's settings, so computers are audited with the free Devseis desktop app for Mac, Windows (beta) and Linux. Download it from the Download page, install it, click Run audit and enter the computer password once. The first run downloads the AI auditor; the audit then takes about 5 to 15 minutes and ends with a PDF report.\n[4] Outdated third-party applications on macOS: Outdated third-party applications on macOS. Why it matters: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Risk if missing: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages. Requirement: Installed third-party applications with a newer version available <= MAX_OUTDATED_APPS (Windows: winget; macOS: app versions; Linux: snap/flatpak and packages). Phones: automatic app updates are on. How to fix on macOS: (1) Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM. How to verify: Re-run the audit; outdated applications should be 0 or fewer. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nQUESTION: Which of my apps are outdated?"}, {"role": "assistant", "content": "I can't see your device from here, so I can't tell you yet. Run an audit with the free Devseis desktop app: it reads the settings, changes nothing, and takes about 5 to 15 minutes [3]. Then ask me again and I will answer from your real results."}], "meta": {"task": "chat", "os": "macos", "status": "noaudit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (windows):\n- Enable real-time protection in Microsoft Defender (on-access protection follows it).\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Compliant\", \"title\": \"On-access protection matches the required setting\", \"finding\": \"Found: on. This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device: does not meet the baseline\", \"finding\": \"The user reports off, but the baseline requires on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Compliant\", \"title\": \"AI transparency to users: meets the baseline\", \"finding\": \"AI transparency to users is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 50\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Screen lock password requirement on Linux: Screen lock password requirement on Linux. Why it matters: Unlocking the screen must require the user's password. Risk if missing: Medium - A screen lock that needs no password does not protect anything. Requirement: A password is required to unlock (macOS: within 5 seconds of the screen locking). How to fix on Linux: (1) Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false. How to verify: Re-run the audit; the password requirement should be Compliant. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Time synchronisation on Linux: Time synchronisation on Linux. Why it matters: Accurate clocks are needed to put events in the right order during an investigation. Risk if missing: Low - Wrong timestamps make incident timelines unreliable. Requirement: The clock is synchronised from a time server. How to fix on Linux: (1) timedatectl set-ntp true (or enable chronyd). How to verify: Re-run the audit; Time synchronisation should be Compliant. Related controls: ISO 27001:2022 A.8.17 (Clock synchronization); GDPR Art. 33 (Notification of a personal data breach to the supervisory authority).\n[3] Password complexity on Linux: Password complexity on Linux. Why it matters: Mixing character types makes passwords harder to guess. Risk if missing: Medium - Simple passwords are cracked quickly by dictionary attacks. Requirement: Complexity enforced (mixed character classes). How to fix on Linux: (1) Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf. How to verify: Re-run the audit; Password complexity should be Compliant. Related controls: ISO 27001:2022 A.5.17 (Authentication information); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] ClamAV malware protection service on Linux: ClamAV malware protection service on Linux. Why it matters: The malware protection service must be running for protection to work. Risk if missing: High - A stopped protection service leaves the device unprotected. Requirement: The platform malware protection service is running. How to fix on Linux: (1) systemctl enable --now clamav-daemon (RHEL: clamd@scan). How to verify: Re-run the audit; the malware protection service should be Compliant. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer (x86-64), Firefox 143; detected by the browser, security settings not checked.\n\nQUESTION: What is the PCI DSS requirement 8.3.6?"}, {"role": "assistant", "content": "I don't have reliable information about that in my knowledge base, so I won't guess. I cover security settings on Windows, macOS, Linux, iPhone and Android, ISO 27001, GDPR and the EU AI Act."}], "meta": {"task": "chat", "os": "any", "status": "unknown", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: type NoSync, service Stopped/Manual\nREQUIRED: NTP or domain time\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (windows):\n- w32tm /config /syncfromflags:manual /manualpeerlist:\"pool.ntp.org\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"Time synchronisation is type NoSync, service Stopped/Manual instead of NTP or domain time. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"w32tm /config /syncfromflags:manual /manualpeerlist:\\\"pool.ntp.org\\\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.5.2 (macos)\nHOSTNAME: HR-MacBook-31\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=15, Non-Compliant=31, NotApplicable=6, Pending=2, Error=2\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 12\n- password.history | Password history | Non-Compliant | not set\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Non-Compliant | not set\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5228\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Non-Compliant | XProtect 5376, 61 days old\n- malware.signature_updates | Malware security updates | Non-Compliant | ConfigDataInstall=0, CriticalUpdateInstall=1\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 180 idle, password delay 300\n- screen_lock.enabled | Screen lock | Non-Compliant | off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after off seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | jdoe\n- user.account_type | Signed-in user account type | Compliant | Standard (mkhan)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Error | not read\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 259 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 37 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.5.2\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Error | not read\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | chromeremotedesktop, filesharing\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.expiration: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.signatures: High\n- malware.signature_updates: Medium\n- screen_lock.timeout: Medium\n- screen_lock.enabled: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=15, assessed=46, percent=33\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Malware protection signatures | High | XProtect 5376, 61 days old\n- admin.unauthorized | Unauthorized employee administrator accounts | High | jdoe\n- vuln.pending_updates | Pending security updates | High | 2 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 259 critical/high\n- gdpr.auto_updates | Automatic security updates | High | check=1 download=1 security=1 macOS-install=0\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\n- vuln.pending_updates: Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked."}, {"role": "assistant", "content": "{\"device\": \"HR-MacBook-31 - macOS 26.5.2\", \"score\": {\"compliant\": 15, \"assessed\": 46, \"percent\": 33}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 31, \"NotApplicable\": 6, \"Pending\": 2, \"Error\": 2}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Malware protection signatures\", \"risk\": \"High\", \"found\": \"XProtect 5376, 61 days old\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"jdoe\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"2 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"259 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic security updates\", \"risk\": \"High\", \"found\": \"check=1 download=1 security=1 macOS-install=0\"}], \"summary\": \"HR-MacBook-31 meets 15 of 46 assessed checks (33%). The most serious gaps are Malware protection signatures, Unauthorized employee administrator accounts and Pending security updates. 2 checks could not be read because the audit was not run as administrator. 2 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Malware protection signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Pending security updates: Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off (no passcode)\nREQUIRED: on\nDETAIL: Without a passcode, iOS data protection is not active.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection (encryption)\", \"finding\": \"Data protection (encryption) is off (no passcode) instead of on. Full-disk encryption keeps data unreadable if the device is lost or stolen. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; no signed-in user to check\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Found: no machine inactivity limit; no signed-in user to check. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: enabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Guest account disabled: does not meet the baseline\", \"finding\": \"Found: enabled. Required: disabled. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker is non-compliant\", \"finding\": \"The device reports off, but the baseline requires on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked: does not meet the baseline\", \"finding\": \"Found: Always. Required: When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect scanning on; Gatekeeper on\nREQUIRED: XProtect and Gatekeeper on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (macos):\n- Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Compliant\", \"title\": \"Apple malware protection service: meets the baseline\", \"finding\": \"Apple malware protection service is set to XProtect scanning on; Gatekeeper on, in line with the required XProtect and Gatekeeper on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"Found: none found. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.7 checked against NVD iOS records. iOS 26.7 (15 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-43688). 32 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in iOS\", \"finding\": \"The device reports 15 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: unified logging active\nREQUIRED: security events logged\nDETAIL: macOS unified log records authentication events; retention is managed by the OS.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (macos):\n- Unified logging is built in; forward security logs to your log platform if long retention is required.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging matches the required setting\", \"finding\": \"The device reports unified logging active, which meets the requirement (security events logged).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: logon: success+failure; account management: success+failure\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging is compliant\", \"finding\": \"Found: logon: success+failure; account management: success+failure. This satisfies the baseline requirement of logon success+failure and account changes audited.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"Found: no. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAI TOOLS FOUND: Ollama\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local AI service listening on port 11434. Local models: C:\\Users\\jdoe\\.ollama\\models (6 GB).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Ollama.\"}"}], "meta": {"task": "ai_classification", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: GPT4All\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: GPT4All. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude Code, ChatGPT, Microsoft Copilot\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: Claude Code, ChatGPT, Microsoft Copilot. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Compliant\", \"title\": \"Data protection impact assessment: meets the baseline\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"]}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: ChatGPT. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"Found: 10. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Date & time set automatically\", \"finding\": \"Found: set by hand. Required: set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: XProtect scanning on; Gatekeeper assessments disabled\nREQUIRED: XProtect and Gatekeeper on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (macos):\n- Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Apple malware protection service: does not meet the baseline\", \"finding\": \"Apple malware protection service is XProtect scanning on; Gatekeeper assessments disabled instead of XProtect and Gatekeeper on. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Pending\", \"title\": \"Google Play Protect is pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (macos):\n- Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption protection\", \"finding\": \"Disk encryption protection is not encrypted instead of fully encrypted. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Let FileVault encryption finish; keep the Mac on power until fdesetup status reports FileVault is On without progress.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Jan\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Found: Jan. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption protection: does not meet the baseline\", \"finding\": \"Found: not encrypted. Required: active, strong cipher. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"Found: assistant screen access on. Required: screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history: does not meet the baseline\", \"finding\": \"Password history is 0 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with pinHistory = 24.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Network Time: Off\nREQUIRED: network time on\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"Time synchronisation is Network Time: Off instead of network time on. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"sudo systemsetup -setusingnetworktime on.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answer ORG_ROPA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: set automatically\nREQUIRED: set automatically\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Date & time set automatically is compliant\", \"finding\": \"Found: set automatically. This satisfies the baseline requirement of set automatically. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.7 checked against NVD iOS records. iOS 26.7 (15 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-43688). 32 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS is non-compliant\", \"finding\": \"Found: 15 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Answer ORG_ROPA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: type NT5DS, service Running/Manual\nREQUIRED: NTP or domain time\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (windows):\n- w32tm /config /syncfromflags:manual /manualpeerlist:\"pool.ntp.org\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation matches the required setting\", \"finding\": \"Time synchronisation is set to type NT5DS, service Running/Manual, in line with the required NTP or domain time.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: auditd not running, journal volatile\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security event logging\", \"finding\": \"The device reports auditd not running, journal volatile, but the baseline requires auditd running. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LUKS1 aes-cbc-essiv:sha256 128-bit key\nREQUIRED: active, strong cipher\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (linux):\n- Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption protection is non-compliant\", \"finding\": \"Disk encryption protection is LUKS1 aes-cbc-essiv:sha256 128-bit key instead of active, strong cipher. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength is non-compliant\", \"finding\": \"The user reports 4-digit code, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iPadOS 27.0\nREQUIRED: >= iOS 18\nDETAIL: Version entered by the user from Settings › General › About.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version matches the required setting\", \"finding\": \"Supported operating system version is set to iPadOS 27.0, in line with the required >= iOS 18. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store is non-compliant\", \"finding\": \"Apps only from the App Store is apps from outside the App Store instead of App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2025-02 (20 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2025-02 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Android security update is non-compliant\", \"finding\": \"Found: security patch 2025-02 (20 months old). Required: <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age is non-compliant\", \"finding\": \"The device reports 0 days, but the baseline requires >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 64 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-08-01  23:27:44\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Security updates installed recently is 64 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAI TOOLS FOUND: Hugging Face models\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /home/jdoe/.cache/huggingface/hub (23G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Hugging Face models.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 58 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 57; snaps: zoom-client\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Outdated third-party applications is 58 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Answer ORG_ROPA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"Found: assistant screen access on. Required: screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: allow\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Error\", \"title\": \"Firewall outbound policy: could not be checked\", \"finding\": \"No evidence was collected for this check (Requires root: run the collector with sudo.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 13.7.8 (macos)\nAI TOOLS FOUND: Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold: does not meet the baseline\", \"finding\": \"The device reports 10, but the baseline requires 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 14\nDETAIL: pwpolicy global account policy\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (macos):\n- Deploy a Passcode configuration profile through your MDM with minLength = 14.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Minimum password length is 10 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile through your MDM with minLength = 14.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nHOSTNAME: sales-dev-97\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=10, Non-Compliant=37, NotApplicable=4, Pending=4, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Non-Compliant | 12\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 90; accounts: jdoe (99999)\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=4, required classes=0\n- lockout.threshold | Account lockout threshold | Non-Compliant | pam_faillock not enabled\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Non-Compliant | no signature database\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: pmartin (1800s)\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: pmartin, mkhan\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Error | not read\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (pmartin)\n- encryption.enabled | Encryption | Non-Compliant | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | not encrypted\n- vuln.pending_updates | Pending security updates | Non-Compliant | 32 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 60 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 7 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | unattended-upgrades installed=1, enabled=0\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 59 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 22.04.5 LTS\n- gdpr.security_logging | Security event logging | Non-Compliant | auditd not running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Non-Compliant | LightDM guest session allowed\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | vnc\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.expiration: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- malware.antivirus: High\n- malware.service: High\n- malware.signature_updates: Medium\n- malware.signatures: High\n- malware.on_access: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- user.account_type: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=10, assessed=47, percent=21\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.service | ClamAV malware protection service | High | not running\n- malware.signatures | Malware signatures | High | no signature database\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now."}, {"role": "assistant", "content": "{\"device\": \"sales-dev-97 - Ubuntu 22.04.5 LTS 22.04\", \"score\": {\"compliant\": 10, \"assessed\": 47, \"percent\": 21}, \"counts\": {\"Compliant\": 10, \"Non-Compliant\": 37, \"NotApplicable\": 4, \"Pending\": 4, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"no signature database\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}], \"summary\": \"sales-dev-97 meets 10 of 47 assessed checks (21%). The most serious gaps are Antivirus, ClamAV malware protection service and Malware signatures. 1 checks could not be read. 4 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Answer the 4 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (pmartin)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"The device reports Administrator (pmartin), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is set to none, in line with the required only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: key protectors: Tpm, RecoveryPassword\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Compliant\", \"title\": \"BitLocker recovery matches the required setting\", \"finding\": \"Found: key protectors: Tpm, RecoveryPassword. This satisfies the baseline requirement of recovery password present.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 142 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.4 checked against NVD iOS records. iOS 26.4 (142 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-28846). 208 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in iOS\", \"finding\": \"The device reports 142 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (passcode set)\nREQUIRED: on\nDETAIL: iOS encrypts the data on the phone with hardware keys once a passcode is set.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Data protection (encryption): meets the baseline\", \"finding\": \"Data protection (encryption) is set to on (passcode set), in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: File History (jdoe).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Current state: date not readable. Confirm the last backup date in: File History (jdoe).\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nHOSTNAME: dev-ws-95\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=30, Non-Compliant=18, NotApplicable=2, Pending=5, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 90; accounts: jdoe (99999)\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=4, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 1 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | tnguyen, jdoe\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (rschmit)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 3 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 20 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 116 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 26.04.1 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Deja Dup (jdoe)\n- gdpr.backup_recent | Recent backup | Non-Compliant | 64 days ago\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | anydesk\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Ollama, Hugging Face models\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Error | not read\n- aiact.risk_classification | AI system risk classification | Pending | Ollama, Hugging Face models\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- malware.on_access: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- usb.write: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- vuln.outdated_apps: Medium\n- gdpr.updates_recent: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.dpia: Medium\n- org.ai_register: Medium\n- org.ai_literacy: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=30, assessed=48, percent=63\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\n- admin.unauthorized | Unauthorized employee administrator accounts | High | tnguyen, jdoe\n- user.account_type | Signed-in user account type | High | Administrator (rschmit)\n- gdpr.updates_recent | Security updates installed recently | High | 116 days ago\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\n- firewall.inbound: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d."}, {"role": "assistant", "content": "{\"device\": \"dev-ws-95 - Ubuntu 26.04.1 LTS 26.04\", \"score\": {\"compliant\": 30, \"assessed\": 48, \"percent\": 63}, \"counts\": {\"Compliant\": 30, \"Non-Compliant\": 18, \"NotApplicable\": 2, \"Pending\": 5, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"tnguyen, jdoe\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (rschmit)\"}, {\"check_id\": \"gdpr.updates_recent\", \"name\": \"Security updates installed recently\", \"risk\": \"High\", \"found\": \"116 days ago\"}], \"summary\": \"dev-ws-95 meets 30 of 48 assessed checks (63%). The most serious gaps are Firewall, Firewall inbound policy and Unauthorized employee administrator accounts. 1 checks could not be read. 5 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Firewall: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\", \"Firewall inbound policy: ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Answer the 3 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"Password expiration is never instead of 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\", \"For a stand-alone device: net accounts /maxpwage:365\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"The device reports anydesk, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities is non-compliant\", \"finding\": \"Found: no. Required: yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: winrm\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is winrm instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 180 idle, password delay 300\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"Screen lock timeout is 180 idle, password delay 300 instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: network time on\nREQUIRED: network time on\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (macos):\n- sudo systemsetup -setusingnetworktime on.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation: meets the baseline\", \"finding\": \"The device reports network time on, which meets the requirement (network time on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: HR-MacBook-35\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=15, Non-Compliant=29, NotApplicable=4, Pending=8, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Non-Compliant | 5\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5368\n- malware.service | Apple malware protection service | Non-Compliant | XProtect scanning on; Gatekeeper assessments disabled\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5213, 22 days old\n- malware.signature_updates | Malware security updates | Non-Compliant | ConfigDataInstall=0, CriticalUpdateInstall=1\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 120 seconds\n- screen_lock.enabled | Screen lock | Non-Compliant | off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after off seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | akaya\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (akaya)\n- encryption.enabled | Encryption | Non-Compliant | FileVault off\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no FileVault\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 265 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 35 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Non-Compliant | Network Time: Off\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Non-Compliant | guest login 1, guest file sharing 0\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Cursor\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Cursor\n- aiact.risk_classification | AI system risk classification | Pending | Cursor\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | not restricted\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.complexity: Medium\n- lockout.observation: Low\n- malware.service: High\n- malware.signature_updates: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.enabled: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.guest_account: Medium\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.ropa: High\n- org.human_oversight: Medium\nSCORE (computed): compliant=15, assessed=44, percent=34\nTOP GAPS (computed, highest risk first):\n- malware.service | Apple malware protection service | High | XProtect scanning on; Gatekeeper assessments disabled\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | akaya\n- user.account_type | Signed-in user account type | High | Administrator (akaya)\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on."}, {"role": "assistant", "content": "{\"device\": \"HR-MacBook-35 - macOS 26.0.1\", \"score\": {\"compliant\": 15, \"assessed\": 44, \"percent\": 34}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 29, \"NotApplicable\": 4, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"Apple malware protection service\", \"risk\": \"High\", \"found\": \"XProtect scanning on; Gatekeeper assessments disabled\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"akaya\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (akaya)\"}], \"summary\": \"HR-MacBook-35 meets 15 of 44 assessed checks (34%). The most serious gaps are Apple malware protection service, Firewall and Firewall inbound policy. 8 items are pending.\", \"next_steps\": [\"Apple malware protection service: Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\", \"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Answer the 6 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: external AI off unless approved\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Pending\", \"title\": \"Apple Intelligence external integrations restricted: decision pending\", \"finding\": \"The user could not find or confirm this setting. Current state: not answered. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-10 Cumulative Update for Windows 11 Version 23H2 for x64-based Systems (KB5047938), Security Intelligence Update for Microsoft Defender Antivirus - KB2267602 (Version 1.478.0).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Pending security updates is 2 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Data processing agreements is no instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Domain firewall: does not meet the baseline\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Device encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (Android 16 encrypts by default)\nREQUIRED: on\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (android):\n- Update to a supported Android version (10 or later encrypts by default) or replace the phone; set a PIN or password screen lock.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Device encryption is compliant\", \"finding\": \"Device encryption is set to on (Android 16 encrypts by default), in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"The device reports partial, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: deny\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Error\", \"title\": \"Firewall inbound policy could not be read\", \"finding\": \"No evidence was collected for this check (Requires root: run the collector with sudo.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Breach response procedure is no instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nDETAIL: AI apps listed by the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (ios):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"AI software inventory is set to no AI tools found, in line with the required inventory taken. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 220 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 18 applications inventoried, 11 recognised (23 common applications covered). Safari 26.1 (19 CVEs, critical, actively exploited; fixed in 26.6.1; e.g. CVE-2025-14174, CVE-2025-43526); Firefox 152.3 (201 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100768, CVE-2026-100779). 109 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 220 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 12 (android)\nHOSTNAME: ops-galaxy-51\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=7, Non-Compliant=17, NotApplicable=2, Pending=4, Error=0\nEVIDENCE: 2 detected by the browser, 20 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Non-Compliant | no screen lock (Swipe/None)\n- screen_lock.timeout | Screen timeout | Non-Compliant | Screen timeout 10 minutes\n- mobile.passcode_strength | Screen lock strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not rooted | Non-Compliant | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Non-Compliant | on\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 12 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-10 (0 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 12\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- mobile.jailbreak_root: High\n- mobile.app_sources: High\n- mobile.usb_debugging: Medium\n- mobile.find_my_device: Medium\n- vuln.outdated_apps: Medium\n- gdpr.auto_updates: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- org.processor_agreements: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=7, assessed=24, percent=29\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- mobile.jailbreak_root | Not rooted | High | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.os_supported | Supported operating system version | High | Android 12\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- mobile.jailbreak_root: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play."}, {"role": "assistant", "content": "{\"device\": \"ops-galaxy-51 - Android 12\", \"score\": {\"compliant\": 7, \"assessed\": 24, \"percent\": 29}, \"counts\": {\"Compliant\": 7, \"Non-Compliant\": 17, \"NotApplicable\": 2, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not rooted\", \"risk\": \"High\", \"found\": \"rooted or bootloader unlocked\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 12\"}], \"summary\": \"ops-galaxy-51 meets 7 of 24 assessed checks (29%). The most serious gaps are Google Play Protect, Not rooted and Install unknown apps blocked. 4 items are pending. 20 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Not rooted: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\", \"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Answer the 4 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 27.0 (ios)\nHOSTNAME: OPS-iPhone-51\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=11, Non-Compliant=6, NotApplicable=3, Pending=9, Error=0\nEVIDENCE: 0 detected by the browser, 21 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 30 seconds\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Pending | not answered\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 27.0\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.lock_screen_previews: Low\nSCORE (computed): compliant=11, assessed=17, percent=65\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- gdpr.backup_configured | iCloud Backup | High | no backup\n- mobile.passcode_strength | Passcode strength | Medium | 4-digit code\n- mobile.find_my_device | Find My iPhone | Medium | off\n- gdpr.time_sync | Date & time set automatically | Low | set by hand\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- gdpr.backup_configured: Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\n- mobile.passcode_strength: Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code."}, {"role": "assistant", "content": "{\"device\": \"OPS-iPhone-51 - iOS 27.0\", \"score\": {\"compliant\": 11, \"assessed\": 17, \"percent\": 65}, \"counts\": {\"Compliant\": 11, \"Non-Compliant\": 6, \"NotApplicable\": 3, \"Pending\": 9, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Passcode strength\", \"risk\": \"Medium\", \"found\": \"4-digit code\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My iPhone\", \"risk\": \"Medium\", \"found\": \"off\"}, {\"check_id\": \"gdpr.time_sync\", \"name\": \"Date & time set automatically\", \"risk\": \"Low\", \"found\": \"set by hand\"}], \"summary\": \"OPS-iPhone-51 meets 11 of 17 assessed checks (65%). The most serious gaps are Apps only from the App Store and iCloud Backup. 9 items are pending. 21 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"iCloud Backup: Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\", \"Passcode strength: Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in bzserv.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Found: date not readable. Confirm the last backup date in bzserv.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: DeepSeek, Meta AI\nREQUIRED: inventory taken\nDETAIL: AI apps listed by the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (ios):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"Found: DeepSeek, Meta AI. This satisfies the baseline requirement of inventory taken. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: unified logging active\nREQUIRED: security events logged\nDETAIL: macOS unified log records authentication events; retention is managed by the OS.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (macos):\n- Unified logging is built in; forward security logs to your log platform if long retention is required.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging is compliant\", \"finding\": \"Security event logging is set to unified logging active, in line with the required security events logged.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answer ORG_DPIA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (pmartin)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"Found: Administrator (pmartin). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 3 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Sequoia 15.7.3-25G782, Command Line Tools for Xcode-26.1, Safari26.1Sequoia-18.6.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Pending security updates is 3 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Where to find AI assistant access restricted on Android: AI assistant access restricted on Android: Can the AI assistant use what is on your screen? Where to look: Gemini app › your profile picture › Settings › 'Screen context' / 'Use screen context' (other assistants have similar options)\n[2] AI assistant access restricted on Android: AI assistant access restricted on Android. Why it matters: Built-in OS AI features can capture screen content or send text to external AI services. Risk if missing: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose. Requirement: OS-level AI features that capture or send user data are disabled by policy. NotApplicable where the OS has none. Phones: external AI integrations (ChatGPT in Apple Intelligence, assistant access to screen content) are off unless approved. How to fix on Android: (1) In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it. How to verify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[3] DeepSeek (AI tool): DeepSeek: Cloud AI assistant, provider DeepSeek (China), data location: Provider cloud in the People's Republic of China. Default EU AI Act risk class for general office use: minimal. GDPR: No EU adequacy decision for China and no business data processing terms; EU data protection authorities have restricted or investigated it. Do not use with work or personal data. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[4] EU AI Act Annex III High-risk AI systems referred to in Article 6(2): EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice. An auditor looks for: No unassessed use of AI for employment decisions, credit scoring or similar Annex III purposes. On endpoints: A general AI assistant becomes high-risk if used, for example, to screen job applicants.\n\nMY AUDIT: Android 14 (android), audited 2026-10-03; 16 of 26 assessed checks compliant; 10 gaps, 4 pending, 0 not read. Gaps: Google Play Protect: Play Protect scanning off; Install unknown apps blocked: unknown apps allowed for at least one app; App updates: automatic app updates off; Automatic system updates: manual; Google backup: no backup; Sensitive notifications hidden when locked: all content shown; Unapproved AI tools (shadow AI): Otter; AI assistant access restricted: assistant screen access on.\n\nQUESTION: How do I fix ai assistant access restricted?"}, {"role": "assistant", "content": "Your audit found AI assistant access restricted: assistant screen access on. Built-in OS AI features can capture screen content or send text to external AI services. To fix it [2]:\n1. In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "android", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Public firewall is non-compliant\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Operating system AI features restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: none built in\nREQUIRED: restricted by policy\nDETAIL: Linux distributions ship no built-in AI assistant.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (linux):\n- No action needed: Linux distributions have no built-in AI assistant.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Operating system AI features restricted: not assessed on this device\", \"finding\": \"Not applicable on this device: Linux distributions ship no built-in AI assistant.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2025-12 (10 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2025-12 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Android security update\", \"finding\": \"The user reports security patch 2025-12 (10 months old), but the baseline requires <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: within limit\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: mkhan\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout is compliant\", \"finding\": \"The device reports within limit, which meets the requirement (<= 300 seconds).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: auditd running, journal persistent\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging: meets the baseline\", \"finding\": \"Security event logging is set to auditd running, journal persistent, in line with the required auditd running.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: root filesystem not encrypted\nREQUIRED: system disk encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (linux):\n- Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Encryption\", \"finding\": \"The device reports root filesystem not encrypted, but the baseline requires system disk encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nHOSTNAME: DESKTOP-5SFXPML\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=52, Non-Compliant=3, NotApplicable=9, Pending=0, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | CrowdStrike Falcon (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | password-protected screen saver on for all signed-in users\n- screen_lock.timeout | Screen lock timeout | Compliant | 300 seconds (machine inactivity limit)\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | DEVSEIS\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (DEVSEIS\\lmuller)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Compliant | key protectors: Tpm, RecoveryPassword\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | AUOptions=4\n- gdpr.updates_recent | Security updates installed recently | Compliant | 20 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26100\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | winrm\n- gdpr.telemetry | Diagnostic data minimised | Compliant | AllowTelemetry=0 (policy)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.remote_access: High\nSCORE (computed): compliant=52, assessed=55, percent=95\nTOP GAPS (computed, highest risk first):\n- gdpr.backup_configured | Backup configured | High | none found\n- gdpr.remote_access | Remote access services restricted | High | winrm\n- gdpr.backup_recent | Recent backup | Medium | no backup\nFIX GUIDANCE (highest-risk gaps):\n- gdpr.backup_configured: Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\n- gdpr.remote_access: Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\n- gdpr.backup_recent: Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent)."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-5SFXPML - Microsoft Windows 11 Pro 10.0.26100\", \"score\": {\"compliant\": 52, \"assessed\": 55, \"percent\": 95}, \"counts\": {\"Compliant\": 52, \"Non-Compliant\": 3, \"NotApplicable\": 9, \"Pending\": 0, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none found\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"winrm\"}, {\"check_id\": \"gdpr.backup_recent\", \"name\": \"Recent backup\", \"risk\": \"Medium\", \"found\": \"no backup\"}], \"summary\": \"DESKTOP-5SFXPML meets 52 of 55 assessed checks (95%). The most serious gaps are Backup configured and Remote access services restricted.\", \"next_steps\": [\"Backup configured: Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\", \"Remote access services restricted: Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\", \"Recent backup: Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 26.5 (ios)\nAI TOOLS FOUND: Gemini\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Gemini.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Domain outbound\", \"finding\": \"Found: block. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Removable media encryption on Windows: Removable media encryption on Windows. Why it matters: Personal data copied to removable media must be encrypted, or copying must be blocked. Risk if missing: Medium - A lost unencrypted USB drive with personal data is a reportable breach. Requirement: Data cannot be written to unencrypted removable drives. How to fix on Windows: (1) Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access. How to verify: Re-run the audit; Removable media encryption should be Compliant. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption); GDPR Art. 34 (Communication of a personal data breach to the data subject).\n[2] USB storage read on Windows: USB storage read on Windows. Why it matters: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter. Risk if missing: Low - The setting differs from the organisation's agreed USB policy. Requirement: Effective USB storage read access equals USB_READ (allow/deny). How to fix on Windows: (1) Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled. How to verify: Re-run the audit; USB storage read should read allow. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.12 (Data leakage prevention); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] USB storage device access on Windows: USB storage device access on Windows. Why it matters: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives. Risk if missing: Low - The device access setting differs from the organisation's agreed USB policy. Requirement: Whether USB mass storage devices can connect at all equals USB_DEVICE_ACCESS (allow/deny): driver, device-installation policy, USBGuard or MDM restriction. How to fix on Windows: (1) To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4. How to verify: Re-run the audit; USB storage device access should read allow. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.12 (Data leakage prevention); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] USB storage write on Windows: USB storage write on Windows. Why it matters: Writing to USB drives is a common way for data to leave the organisation, often unencrypted. Risk if missing: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost. Requirement: Effective USB storage write access equals USB_WRITE (allow/deny). How to fix on Windows: (1) Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker'). How to verify: Re-run the audit; USB storage write should read deny. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.12 (Data leakage prevention); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Windows 10 or 11 (windows) on a Windows PC (x86-64), Edge 141; detected by the browser, security settings not checked.\n\nQUESTION: How do I fix usb storage write on Windows?"}, {"role": "assistant", "content": "Writing to USB drives is a common way for data to leave the organisation, often unencrypted. To fix it on Windows [4]:\n1. Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\n\nTo check it worked: Re-run the audit; USB storage write should read deny."}], "meta": {"task": "chat", "os": "windows", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: FullyDecrypted (0%)\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"Encryption is FullyDecrypted (0%) instead of fully encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAI TOOLS FOUND: Microsoft Copilot, Hugging Face models\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: C:\\Users\\jdoe\\.cache\\huggingface\\hub (27 GB).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Microsoft Copilot: Cloud AI assistant; provider Microsoft (US); data: Microsoft cloud; enterprise plans keep data under Microsoft's commercial data protection terms; default risk minimal; GDPR: Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Microsoft Copilot\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Microsoft Copilot, Hugging Face models.\"}"}], "meta": {"task": "ai_classification", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Ubuntu 20.04.6 LTS\nREQUIRED: Ubuntu 22.04 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"The device reports Ubuntu 20.04.6 LTS, but the baseline requires Ubuntu 22.04 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"AI transparency to users is partial instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"Found: disabled. Required: enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"ClamAV malware protection service is non-compliant\", \"finding\": \"ClamAV malware protection service is not running instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (windows):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not assessed on this device\", \"finding\": \"This check does not apply: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Public firewall is non-compliant\", \"finding\": \"Public firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an enabled administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"The device reports none of the approved accounts is an enabled administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"Found: off. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: DeepSeek, Meta AI\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: DeepSeek, Meta AI. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 8\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Minimum password length is 8 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: macOS Sequoia 15.9.3-25G188, Command Line Tools for Xcode-26.1.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates: does not meet the baseline\", \"finding\": \"Found: 2 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Compliant\", \"title\": \"Human oversight for high-risk AI: meets the baseline\", \"finding\": \"Human oversight for high-risk AI is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: disabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled matches the required setting\", \"finding\": \"Found: disabled. This satisfies the baseline requirement of disabled.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Bitdefender Endpoint Security (on, up to date); Windows Defender (off, up to date)\nREQUIRED: enabled and up to date\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (windows):\n- Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus: meets the baseline\", \"finding\": \"Antivirus is set to Bitdefender Endpoint Security (on, up to date); Windows Defender (off, up to date), in line with the required enabled and up to date.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: root filesystem not encrypted\nREQUIRED: system disk encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (linux):\n- Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"Found: root filesystem not encrypted. Required: system disk encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Compliant\", \"title\": \"Public inbound: meets the baseline\", \"finding\": \"The device reports block, which meets the requirement (block).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (asmith)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"Found: Administrator (asmith). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: minclass=4, required classes=0\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Compliant\", \"title\": \"Password complexity is compliant\", \"finding\": \"Password complexity is set to minclass=4, required classes=0, in line with the required 3 or more character classes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Google backup is non-compliant\", \"finding\": \"Google backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Microsoft Windows 11 Pro build 22621\nREQUIRED: build 22631 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"The device reports Microsoft Windows 11 Pro build 22621, but the baseline requires build 22631 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: success; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging is non-compliant\", \"finding\": \"Found: logon: success; account management: none. Required: logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no recovery key\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption recovery\", \"finding\": \"The device reports no recovery key, but the baseline requires recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Red Hat Enterprise Linux 8.10 (Ootpa)\nREQUIRED: RHEL 9 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version: does not meet the baseline\", \"finding\": \"The device reports Red Hat Enterprise Linux 8.10 (Ootpa), but the baseline requires RHEL 9 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: tzdata, systemd, libssl3t64. Package lists are 14 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Pending security updates is 5 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Compliant\", \"title\": \"Data protection impact assessment matches the required setting\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: rdp\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is rdp instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: XProtect scanning on; Gatekeeper assessments disabled\nREQUIRED: XProtect and Gatekeeper on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (macos):\n- Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Apple malware protection service: does not meet the baseline\", \"finding\": \"The device reports XProtect scanning on; Gatekeeper assessments disabled, but the baseline requires XProtect and Gatekeeper on. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"The device reports Claude Code, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"Found: whoopsie error reporting on. Required: usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"Found: whoopsie error reporting on. Required: usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MIN_DAYS 0\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (linux):\n- Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age is non-compliant\", \"finding\": \"Found: PASS_MIN_DAYS 0. Required: >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"Lockout observation period is 10 minutes instead of >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: screen saver does not require sign-in for: CONTOSO\\akaya\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"Found: screen saver does not require sign-in for: CONTOSO\\\\akaya. Required: sign-in required to unlock. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 26.0.1 (macos)\nAI TOOLS FOUND: Claude\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: tnguyen (3600s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: tnguyen\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"Found: too long for: tnguyen (3600s). Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no recovery key\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery: does not meet the baseline\", \"finding\": \"Disk encryption recovery is no recovery key instead of recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Google backup: does not meet the baseline\", \"finding\": \"Found: no backup. Required: backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: tnguyen\nREQUIRED: password required\nDETAIL: Users checked: tnguyen\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"Screen lock password requirement is not required for: tnguyen instead of password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 221 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 22 applications inventoried, 13 recognised (31 common applications covered). Notepad++ (64-bit x64) 8.4.5 (5 CVEs, high, actively exploited; fixed in 8.9.6.1; e.g. CVE-2023-40031, CVE-2025-15556); Mozilla Firefox (x64 en-US) 152.0.6 (201 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100768, CVE-2026-100779); Zoom Workplace 6.4.5 (9 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-46788, CVE-2025-49460); Microsoft Teams 1.6.00.26474 (6 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330). 56 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Known vulnerabilities in installed software is 221 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: unified logging active\nREQUIRED: security events logged\nDETAIL: macOS unified log records authentication events; retention is managed by the OS.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (macos):\n- Unified logging is built in; forward security logs to your log platform if long retention is required.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging is compliant\", \"finding\": \"Security event logging is set to unified logging active, in line with the required security events logged.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: required\nREQUIRED: password required\nDETAIL: Users checked: rschmit\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement matches the required setting\", \"finding\": \"The device reports required, which meets the requirement (password required).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 14 (android)\nAI TOOLS FOUND: ChatGPT\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Perplexity, ChatGPT, Claude\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (macos):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"Found: Perplexity, ChatGPT, Claude. This satisfies the baseline requirement of inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound is non-compliant\", \"finding\": \"The device reports block, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 15.7.2 (macos)\nAI TOOLS FOUND: LM Studio, Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- LM Studio: Local model runner; provider Element Labs (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences and whether its local server is exposed.\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"LM Studio\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences and whether its local server is exposed.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: LM Studio, Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: popularity-contest on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"Found: popularity-contest on. Required: usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 16 (android)\nAI TOOLS FOUND: Gemini\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Gemini.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: NTP service no, synchronised no\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"Found: NTP service no, synchronised no. Required: time synchronised. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"timedatectl set-ntp true (or enable chronyd).\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: jdoe. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public firewall\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall\", \"finding\": \"The device reports no active firewall, but the baseline requires active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 29 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: libxml2, openssl, sudo. Package lists are 2 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Found: 29 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (akaya)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"Signed-in user account type is Administrator (akaya) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (DESKTOP-R2S2I1D\\mkhan)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Signed-in user account type is Administrator (DESKTOP-R2S2I1D\\\\mkhan) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Pending\", \"title\": \"Find My Device: decision pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nHOSTNAME: DESKTOP-1V6Y5K3\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=21, Non-Compliant=38, NotApplicable=4, Pending=1, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 0\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 0\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Compliant | on\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Non-Compliant | off\n- malware.signatures | Security intelligence | Non-Compliant | 24 days old (version 1.430.686.0)\n- firewall.domain.enabled | Domain firewall | Non-Compliant | off\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Non-Compliant | block\n- firewall.private.enabled | Private firewall | Non-Compliant | off\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Non-Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Non-Compliant | block\n- screen_lock.enabled | Screen lock | Non-Compliant | no machine inactivity limit; no signed-in user to check\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | screen saver does not require sign-in for: DEVSEIS\\pmartin\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an enabled administrator\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-1V6Y5K3\\ewagner\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (DEVSEIS\\pmartin)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, XtsAes128\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 6 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 314 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 13 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 10 Pro build 19045\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Non-Compliant | type NoSync, service Stopped/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Non-Compliant | enabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | anydesk\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- malware.tamper: Medium\n- malware.signatures: High\n- firewall.domain.enabled: High\n- firewall.domain.inbound: High\n- firewall.domain.outbound: Low\n- firewall.private.enabled: High\n- firewall.private.inbound: High\n- firewall.private.outbound: Low\n- firewall.public.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.ropa: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=21, assessed=59, percent=36\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Security intelligence | High | 24 days old (version 1.430.686.0)\n- firewall.domain.enabled | Domain firewall | High | off\n- firewall.domain.inbound | Domain inbound | High | allow\n- firewall.private.enabled | Private firewall | High | off\n- firewall.private.inbound | Private inbound | High | allow\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: Run Update-MpSignature and check that the device can reach Windows Update or your update server.\n- firewall.domain.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-1V6Y5K3 - Microsoft Windows 10 Pro 10.0.19045\", \"score\": {\"compliant\": 21, \"assessed\": 59, \"percent\": 36}, \"counts\": {\"Compliant\": 21, \"Non-Compliant\": 38, \"NotApplicable\": 4, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Security intelligence\", \"risk\": \"High\", \"found\": \"24 days old (version 1.430.686.0)\"}, {\"check_id\": \"firewall.domain.enabled\", \"name\": \"Domain firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.private.enabled\", \"name\": \"Private firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}], \"summary\": \"DESKTOP-1V6Y5K3 meets 21 of 59 assessed checks (36%). The most serious gaps are Security intelligence, Domain firewall and Domain inbound. 1 items are pending.\", \"next_steps\": [\"Security intelligence: Run Update-MpSignature and check that the device can reach Windows Update or your update server.\", \"Domain firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rschmit, tnguyen\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: rschmit, tnguyen. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Recent backup on Linux: Recent backup on Linux. Why it matters: A configured backup only helps if it has actually run recently. Risk if missing: Medium - Recent work would be lost if the device failed today. Requirement: Last backup within BACKUP_MAX_AGE_DAYS days (Pending when the date cannot be read). How to fix on Linux: (1) Run the backup job now and check its timer and logs. How to verify: Re-run the audit; the last backup should be within 7 days. Related controls: ISO 27001:2022 A.8.13 (Information backup); GDPR Art. 32(1)(c) (Security of processing: restore availability and access after an incident).\n[2] Diagnostic data minimised on Linux: Diagnostic data minimised on Linux. Why it matters: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Risk if missing: Low - More personal data than necessary leaves the organisation. Requirement: OS diagnostic/usage data sharing is at the minimum level. How to fix on Linux: (1) Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie). How to verify: Re-run the audit; Diagnostic data minimised should be Compliant. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default).\n[3] EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: EU AI Act Art. 26(6) (Deployers: keep automatically generated logs for at least six months): Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise. An auditor looks for: Log retention of at least six months for high-risk AI systems. On endpoints: Only applies if a detected AI tool is used in a high-risk way.\n[4] AI usage log retention on Linux: AI usage log retention on Linux. Why it matters: Deployers of high-risk AI systems must keep the system's logs for at least six months. Risk if missing: Medium - Missing logs prevent monitoring and investigation of high-risk AI use. Requirement: Logs kept >= 6 months if a high-risk AI system is used. Pending until classification; NotApplicable when no AI tool is found. How to fix on Linux: (1) If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months. How to verify: Log retention is documented for each high-risk AI system. Related controls: ISO 27001:2022 A.8.15 (Logging); EU AI Act Art. 26(6) (Deployers: keep automatically generated logs for at least six months).\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer, Firefox 143; detected by the browser, security settings not checked.\n\nQUESTION: How do I check that ai usage log retention is set correctly on Linux?"}, {"role": "assistant", "content": "Log retention is documented for each high-risk AI system. [4]"}], "meta": {"task": "chat", "os": "linux", "status": "verify", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 14\nDETAIL: pwpolicy global account policy\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (macos):\n- Deploy a Passcode configuration profile through your MDM with minLength = 14.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password length\", \"finding\": \"Found: not set. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile through your MDM with minLength = 14.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: screen saver does not require sign-in for: CONTOSO\\tnguyen\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"Found: screen saver does not require sign-in for: CONTOSO\\\\tnguyen. Required: sign-in required to unlock. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: protection On, XtsAes128\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"BitLocker protection: meets the baseline\", \"finding\": \"BitLocker protection is set to protection On, XtsAes128, in line with the required protection on, AES.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Found: USB storage writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: only: none\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN).\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Error\", \"title\": \"Remote access services restricted could not be read\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: ChatGPT\nREQUIRED: inventory taken\nDETAIL: AI apps listed by the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (ios):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"The user reports ChatGPT, which meets the requirement (inventory taken). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (mkhan)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"The device reports Administrator (mkhan), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery: does not meet the baseline\", \"finding\": \"Disk encryption recovery is 1 key slots, 0 unlock tokens instead of recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: lmuller (1800s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: lmuller\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: too long for: lmuller (1800s). Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Data protection impact assessment is partial instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Encryption on Windows: Encryption on Windows. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: The system disk is fully encrypted (BitLocker / LUKS / FileVault). Phones: iOS data protection is on when a passcode is set; Android 10+ encrypts by default. How to fix on Windows: (1) Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[2] BitLocker protection on Windows: BitLocker protection on Windows. Why it matters: Encryption only protects the data when it is active, fully applied and uses a strong cipher. Risk if missing: High - Suspended or partial encryption leaves data readable on a lost device. Requirement: Encryption protection is active (not suspended or in progress) with a strong cipher. How to fix on Windows: (1) Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method. How to verify: Re-run the audit; Disk encryption protection should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[3] Removable media encryption on Windows: Removable media encryption on Windows. Why it matters: Personal data copied to removable media must be encrypted, or copying must be blocked. Risk if missing: Medium - A lost unencrypted USB drive with personal data is a reportable breach. Requirement: Data cannot be written to unencrypted removable drives. How to fix on Windows: (1) Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access. How to verify: Re-run the audit; Removable media encryption should be Compliant. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption); GDPR Art. 34 (Communication of a personal data breach to the data subject).\n[4] BitLocker on Windows: BitLocker on Windows. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: BitLocker is turned on for the system drive. How to fix on Windows: (1) Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n\nMY AUDIT: none\n\nQUESTION: How do I check that encryption is set correctly on Windows?"}, {"role": "assistant", "content": "Re-run the audit; Encryption should be Compliant. [1]"}], "meta": {"task": "chat", "os": "windows", "status": "verify", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: enabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Guest account disabled\", \"finding\": \"Found: enabled. Required: disabled. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Unauthorized employee administrator accounts is jdoe instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup is non-compliant\", \"finding\": \"The user reports no backup, but the baseline requires backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Install unknown apps blocked: does not meet the baseline\", \"finding\": \"Install unknown apps blocked is unknown apps allowed for at least one app instead of all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public firewall\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Sensitive notifications hidden when locked: does not meet the baseline\", \"finding\": \"Found: all content shown. Required: sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training: meets the baseline\", \"finding\": \"AI literacy training is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAI TOOLS FOUND: Jan\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Jan: Local model runner; provider Open source (runs locally); data: On the device unless a cloud model is connected; default risk minimal; GDPR: Check whether remote API providers are configured; if so, the cloud provider rules apply.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Jan\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Check whether remote API providers are configured; if so, the cloud provider rules apply.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Jan.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iCloud Backup\", \"finding\": \"iCloud Backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-09 Cumulative Update for Windows 11 Version 24H2 for x64-based Systems (KB5043715).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates: does not meet the baseline\", \"finding\": \"The device reports 1 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 17.7 (ios)\nHOSTNAME: FIN-iPhone-52\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=5, Non-Compliant=13, NotApplicable=2, Pending=9, Error=0\nEVIDENCE: 3 detected by the browser, 18 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Non-Compliant | no passcode\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Non-Compliant | off (no passcode)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 17.7 (update to 17.7.2 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 317 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Pending | not answered\n- gdpr.os_supported | Supported operating system version | Non-Compliant | iOS 17.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- encryption.enabled: High\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.lock_screen_previews: Low\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=5, assessed=18, percent=28\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- encryption.enabled | Data protection (encryption) | High | off (no passcode)\n- vuln.pending_updates | iOS security updates | High | iOS 17.7 (update to 17.7.2 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 317 critical/high\n- gdpr.os_supported | Supported operating system version | High | iOS 17.7\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- encryption.enabled: Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered."}, {"role": "assistant", "content": "{\"device\": \"FIN-iPhone-52 - iOS 17.7\", \"score\": {\"compliant\": 5, \"assessed\": 18, \"percent\": 28}, \"counts\": {\"Compliant\": 5, \"Non-Compliant\": 13, \"NotApplicable\": 2, \"Pending\": 9, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Data protection (encryption)\", \"risk\": \"High\", \"found\": \"off (no passcode)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 17.7 (update to 17.7.2 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"317 critical/high\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"iOS 17.7\"}], \"summary\": \"FIN-iPhone-52 meets 5 of 18 assessed checks (28%). The most serious gaps are Apps only from the App Store, Data protection (encryption) and iOS security updates. 9 items are pending. 18 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Data protection (encryption): Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 142 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-05-16  19:13:30\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security updates installed recently\", \"finding\": \"Security updates installed recently is 142 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery: does not meet the baseline\", \"finding\": \"Found: not encrypted. Required: recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 180 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration is compliant\", \"finding\": \"Password expiration is set to 180 days, in line with the required 1-365 days.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Recall feature Disabled, not disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Windows Recall / Copilot data analysis disabled is non-compliant\", \"finding\": \"The device reports Recall feature Disabled, not disabled by policy, but the baseline requires disabled by policy. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: set automatically\nREQUIRED: set automatically\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Date & time set automatically is compliant\", \"finding\": \"Found: set automatically. This satisfies the baseline requirement of set automatically. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Compliant\", \"title\": \"Public inbound: meets the baseline\", \"finding\": \"Public inbound is set to block, in line with the required block.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"On-access malware protection is not running instead of on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"AI assistant access restricted is assistant screen access on instead of screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 15 (android)\nAI TOOLS FOUND: Claude, Meta AI, Replika\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\n- Meta AI: Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps); provider Meta (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\n- Replika: AI companion chatbot; provider Luka (US); data: Provider cloud outside the EEA; default risk limited; GDPR: Consumer companion app that collects sensitive personal conversations and has been subject to EU data protection enforcement; not suitable for work use.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Meta AI\", \"kind\": \"Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Replika\", \"kind\": \"AI companion chatbot\", \"approved\": false, \"default_risk_class\": \"limited\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer companion app that collects sensitive personal conversations and has been subject to EU data protection enforcement; not suitable for work use.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude, Meta AI, Replika.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: firewalld active\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Compliant\", \"title\": \"Firewall matches the required setting\", \"finding\": \"The device reports firewalld active, which meets the requirement (active).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.7 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates: does not meet the baseline\", \"finding\": \"The device reports iOS 26.7 (update to 26.7.1 available), but the baseline requires latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Diagnostic data minimised is whoopsie error reporting on instead of usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: itadmin\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Compliant\", \"title\": \"Approved IT administrator: meets the baseline\", \"finding\": \"The device reports itadmin, which meets the requirement (itadmin).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"The device reports partial, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register is non-compliant\", \"finding\": \"AI register is no instead of yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Cursor\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The device reports Cursor, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates is non-compliant\", \"finding\": \"Found: manual. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Found: partial. Required: yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Block or allow external storage with your MDM (Restrictions / device control) or an endpoint DLP product.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"USB storage device access is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamav-freshclam\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Compliant\", \"title\": \"Malware signature updates matches the required setting\", \"finding\": \"The device reports clamav-freshclam, which meets the requirement (automatic).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Gemini\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The user reports Gemini, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAI TOOLS FOUND: GPT4All, Ollama, Alpaca (Ollama client)\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /home/jdoe/.ollama/models (50G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- GPT4All: Local model runner; provider Nomic AI (runs locally); data: On the device unless optional data sharing is enabled; default risk minimal; GDPR: Keep the optional data lake sharing switched off.\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\n- Alpaca (Ollama client): Desktop client for local models; provider Open source (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"GPT4All\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Keep the optional data lake sharing switched off.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Alpaca (Ollama client)\", \"kind\": \"Desktop client for local models\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: GPT4All, Ollama, Alpaca (Ollama client).\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude, Cursor, Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: Claude, Cursor, Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Microsoft Copilot\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Found: Microsoft Copilot. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery: does not meet the baseline\", \"finding\": \"The device reports 1 key slots, 0 unlock tokens, but the baseline requires recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"The device reports no, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (CONTOSO\\akaya)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Found: Administrator (CONTOSO\\\\akaya). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is ChatGPT instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Restrict external storage with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read matches the required setting\", \"finding\": \"USB storage read is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound is non-compliant\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"The device reports USB storage writable without enforced encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 50 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 21 applications inventoried, 10 recognised (31 common applications covered). Adobe Acrobat Reader (64-bit) 25.001.21288 (38 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2026-34621, CVE-2026-34622); Zoom Workplace 6.4.10 (9 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-46788, CVE-2025-49460); Microsoft Edge 152.0.4191.66 (3 CVEs, high; fixed in 153.0.4234.46; e.g. CVE-2026-69486, CVE-2026-85893). 26 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"The device reports 50 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 60 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration: meets the baseline\", \"finding\": \"The device reports 60 days, which meets the requirement (1-365 days).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: success; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging: does not meet the baseline\", \"finding\": \"The device reports logon: success; account management: none, but the baseline requires logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (akaya)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"Signed-in user account type is Administrator (akaya) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Hugging Face models\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: ChatGPT, Hugging Face models. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"Found: 10 minutes. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Compliant\", \"title\": \"Human oversight for high-risk AI matches the required setting\", \"finding\": \"Human oversight for high-risk AI is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 223 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 24 applications inventoried, 13 recognised (31 common applications covered). Adobe Acrobat Reader (64-bit) 25.001.20997 (40 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2026-27220, CVE-2026-27278); Notepad++ (64-bit x64) 8.4.5 (5 CVEs, high, actively exploited; fixed in 8.9.6.1; e.g. CVE-2023-40031, CVE-2025-15556); Mozilla Firefox (x64 en-US) 153.0.0 (154 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100756, CVE-2026-100767); VLC media player 2.1.6 (16 CVEs, critical; fixed in 3.0.20; e.g. CVE-2016-5108, CVE-2017-17670); Zoom Workplace 6.6.11 (2 CVEs, critical; fixed in 7.0.5; e.g. CVE-2026-53410, CVE-2026-53412); and 1 more. 72 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"The device reports 223 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iPadOS 18.7.9 (ios)\nHOSTNAME: HR-iPad-41\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=12, Non-Compliant=7, NotApplicable=2, Pending=8, Error=0\nEVIDENCE: 2 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 3 minutes\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Non-Compliant | jailbroken\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iPadOS 18.7.9 (update to 18.7.10 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 160 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | download only\n- gdpr.os_supported | Supported operating system version | Compliant | iPadOS 18.7.9\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | iCloud Backup | Compliant | backed up to a computer\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | ChatGPT extension off\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.app_sources: High\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.time_sync: Low\nSCORE (computed): compliant=12, assessed=19, percent=63\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not jailbroken | High | jailbroken\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- vuln.pending_updates | iOS security updates | High | iPadOS 18.7.9 (update to 18.7.10 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 160 critical/high\n- gdpr.auto_updates | Automatic iOS updates | High | download only\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered."}, {"role": "assistant", "content": "{\"device\": \"HR-iPad-41 - iPadOS 18.7.9\", \"score\": {\"compliant\": 12, \"assessed\": 19, \"percent\": 63}, \"counts\": {\"Compliant\": 12, \"Non-Compliant\": 7, \"NotApplicable\": 2, \"Pending\": 8, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not jailbroken\", \"risk\": \"High\", \"found\": \"jailbroken\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iPadOS 18.7.9 (update to 18.7.10 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"160 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"download only\"}], \"summary\": \"HR-iPad-41 meets 12 of 19 assessed checks (63%). The most serious gaps are Not jailbroken, Apps only from the App Store and iOS security updates. 8 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not jailbroken: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\", \"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: security responses and system data files install automatically\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Compliant\", \"title\": \"Malware security updates: meets the baseline\", \"finding\": \"Malware security updates is set to security responses and system data files install automatically, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: automatic app updates on\nREQUIRED: automatic app updates on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Compliant\", \"title\": \"App updates: meets the baseline\", \"finding\": \"App updates is set to automatic app updates on, in line with the required automatic app updates on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 15 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period: meets the baseline\", \"finding\": \"Found: 15 minutes. This satisfies the baseline requirement of >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Breach response procedure is no instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 15 (android)\nHOSTNAME: fin-pixel-15\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=6, Non-Compliant=14, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Non-Compliant | Screen timeout 30 minutes\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | swipe or none\n- mobile.jailbreak_root | Not rooted | Non-Compliant | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 15 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2026-05 (5 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Compliant | Android 15\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | DeepSeek, Grok\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | DeepSeek, Grok\n- aiact.risk_classification | AI system risk classification | Pending | DeepSeek, Grok\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=6, assessed=20, percent=30\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- mobile.jailbreak_root | Not rooted | High | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- vuln.pending_updates | Android security update | High | security patch 2026-05 (5 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- mobile.jailbreak_root: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play."}, {"role": "assistant", "content": "{\"device\": \"fin-pixel-15 - Android 15\", \"score\": {\"compliant\": 6, \"assessed\": 20, \"percent\": 30}, \"counts\": {\"Compliant\": 6, \"Non-Compliant\": 14, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not rooted\", \"risk\": \"High\", \"found\": \"rooted or bootloader unlocked\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2026-05 (5 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}], \"summary\": \"fin-pixel-15 meets 6 of 20 assessed checks (30%). The most serious gaps are Google Play Protect, Not rooted and Install unknown apps blocked. 10 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Not rooted: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\", \"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Domain firewall is non-compliant\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: type NoSync, service Stopped/Manual\nREQUIRED: NTP or domain time\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (windows):\n- w32tm /config /syncfromflags:manual /manualpeerlist:\"pool.ntp.org\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation: does not meet the baseline\", \"finding\": \"The device reports type NoSync, service Stopped/Manual, but the baseline requires NTP or domain time. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"w32tm /config /syncfromflags:manual /manualpeerlist:\\\"pool.ntp.org\\\" /update (domain members sync from the domain automatically); make sure the Windows Time service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Google Play Protect is non-compliant\", \"finding\": \"Found: Play Protect scanning off. Required: on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"The device reports none of the approved accounts is an administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"Minimum password age is 0 days instead of >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (windows):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not applicable\", \"finding\": \"Not applicable on this device: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: backup on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (android):\n- Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Pending\", \"title\": \"Google backup: decision pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: machine inactivity limit locks the session\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement matches the required setting\", \"finding\": \"Found: machine inactivity limit locks the session. This satisfies the baseline requirement of sign-in required to unlock.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an enabled administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator is non-compliant\", \"finding\": \"Found: none of the approved accounts is an enabled administrator. Required: itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 349 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 18 applications inventoried, 13 recognised (31 common applications covered). Mozilla Firefox (x64 en-US) 152.3 (201 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100768, CVE-2026-100779); Google Chrome 153.0.8010.52 (137 CVEs, critical; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301); Zoom Workplace 6.3.10 (11 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-30663, CVE-2025-30664). 198 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 349 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"Found: not encrypted. Required: recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall: does not meet the baseline\", \"finding\": \"Found: no active firewall. Required: active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode on\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Found: firewall off, stealth mode on. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic security updates\", \"finding\": \"Automatic security updates is check=1 download=1 security=1 macOS-install=0 instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] ISO 27001:2022 A.8.24 Use of cryptography: ISO 27001:2022 A.8.24 (Use of cryptography): Define and apply rules for using cryptography, including key management. An auditor looks for: Full-disk encryption with strong algorithms; Recovery keys stored safely. On endpoints: Disk encryption protects data on lost or stolen devices.\n[2] BitLocker on Windows: BitLocker on Windows. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: BitLocker is turned on for the system drive. How to fix on Windows: (1) Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[3] GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: GDPR Art. 32(1)(c) (Security of processing: restore availability and access after an incident): Be able to restore access to personal data in time after a physical or technical incident. An auditor looks for: Working backups; Recovery keys available. On endpoints: Backups and disk-encryption recovery keys make restoration possible.\n[4] BitLocker recovery on Windows: BitLocker recovery on Windows. Why it matters: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes. Risk if missing: Medium - Without a recovery key, an encrypted disk can become permanently unreadable. Requirement: A recovery key or second unlock method exists. How to fix on Windows: (1) Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory. How to verify: Re-run the audit; Disk encryption recovery should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.13 (Information backup); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption); GDPR Art. 32(1)(c) (Security of processing: restore availability and access after an incident).\n\nMY AUDIT: none\n\nQUESTION: How do I rotate keys in HashiCorp Vault?"}, {"role": "assistant", "content": "I don't have reliable information about that in my knowledge base, so I won't guess. I cover security settings on Windows, macOS, Linux, iPhone and Android, ISO 27001, GDPR and the EU AI Act."}], "meta": {"task": "chat", "os": "any", "status": "unknown", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] ChatGPT (AI tool): ChatGPT: Cloud AI assistant, provider OpenAI (US), data location: Provider cloud outside the EEA unless an EU data residency plan is used. Default EU AI Act risk class for general office use: minimal. GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[2] ISO 27001:2022 A.5.24 Information security incident management planning and preparation: ISO 27001:2022 A.5.24 (Information security incident management planning and preparation): Plan in advance how security incidents are reported, handled and communicated. An auditor looks for: A documented incident procedure with roles; Contact points and escalation paths. On endpoints: Endpoint logs and alerts feed the incident process.\n[3] Windsurf (AI tool): Windsurf: AI code editor, provider US-based provider, data location: Code sent to the provider's cloud. Default EU AI Act risk class for general office use: minimal. GDPR: Use a business plan with a data processing agreement and zero-retention settings. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[4] Claude (AI tool): Claude: Cloud AI assistant, provider Anthropic (US), data location: Provider cloud outside the EEA. Default EU AI Act risk class for general office use: minimal. GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44). Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n\nMY AUDIT: none\n\nQUESTION: Plan a weekend trip to Paris."}, {"role": "assistant", "content": "I can only help with device security, ISO 27001, GDPR and the EU AI Act, so I can't answer that. Ask me, for example, how to turn on disk encryption, whether an AI tool is OK for work, or run an audit of this device."}], "meta": {"task": "chat", "os": "any", "status": "oos", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] How to install on a Mac: Open the downloaded .dmg file and drag Devseis Endpoint Auditor to Applications. The first time, macOS says it could not verify the app because it is not yet notarised: click Done, then open System Settings, Privacy & Security, and click Open Anyway. After that it opens normally. It needs macOS 13 or newer.\n[2] How to run an audit of a computer: A web page cannot read a computer's settings, so computers are audited with the free Devseis desktop app for Mac, Windows (beta) and Linux. Download it from the Download page, install it, click Run audit and enter the computer password once. The first run downloads the AI auditor; the audit then takes about 5 to 15 minutes and ends with a PDF report.\n[3] Password expiration on macOS: Password expiration on macOS. Why it matters: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful. Risk if missing: Low - A leaked password can stay valid indefinitely. Requirement: Passwords expire after 1..PASSWORD_MAX_AGE_DAYS days (0 in the baseline = expiry not required). How to fix on macOS: (1) Deploy a Passcode configuration profile with maxPINAgeInDays = 365. How to verify: Re-run the audit; Password expiration should be between 1 and 365 days. Related controls: ISO 27001:2022 A.5.17 (Authentication information); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Does the auditor need internet?: Only the first time, to download the AI auditor and the vulnerability list. After that audits and chat work offline. Pending security updates can only be checked online because they ask the vendor.\n\nMY AUDIT: macOS 15.7.2 (macos), audited 2026-10-04; 42 of 51 assessed checks compliant; 9 gaps, 0 pending, 0 not read. Gaps: Password expiration: never; Lockout observation period: 10 minutes; Firewall inbound policy: firewall on, stealth mode off; USB storage device access: deny; USB storage read: deny; Unauthorized employee administrator accounts: pmartin; Outdated third-party applications: 48 outdated; Backup configured: none.\n\nQUESTION: What should I fix first?"}, {"role": "assistant", "content": "Your audit found 9 gaps. Start with Password expiration (found: never). The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful. To fix it [3]:\n1. Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\n\nThen look at: Lockout observation period, Firewall inbound policy, USB storage device access."}], "meta": {"task": "chat", "os": "macos", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: protection On, XtsAes128\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Compliant\", \"title\": \"BitLocker protection is compliant\", \"finding\": \"BitLocker protection is set to protection On, XtsAes128, in line with the required protection on, AES.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: filesharing\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"Found: filesharing. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: none\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Error\", \"title\": \"Unauthorized employee administrator accounts: evidence missing\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Gemini\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"The user reports Gemini, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"The device reports 1 key slots, 0 unlock tokens, but the baseline requires recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no screen lock (Swipe/None)\nREQUIRED: PIN, password or pattern\nDETAIL: Anyone holding the phone can open it.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"The user reports no screen lock (Swipe/None), but the baseline requires PIN, password or pattern. An idle device should lock itself so nobody can use it while the owner is away. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: auditd running, journal volatile\nREQUIRED: auditd running\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (linux):\n- Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf).\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging: meets the baseline\", \"finding\": \"Found: auditd running, journal volatile. This satisfies the baseline requirement of auditd running.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (sali)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Found: Administrator (sali). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: no such setting\nREQUIRED: >= 1 days\nDETAIL: macOS account policies have no minimum password age.\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (macos):\n- macOS has no minimum password age setting; rely on password history instead.\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"NotApplicable\", \"title\": \"Minimum password age: not assessed on this device\", \"finding\": \"Not applicable on this device: macOS account policies have no minimum password age.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Sensitive notifications hidden when locked: does not meet the baseline\", \"finding\": \"The user reports all content shown, but the baseline requires sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI) matches the required setting\", \"finding\": \"The user reports none, which meets the requirement (only: none). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-10 Cumulative Update for Windows 11 Version 25H2 for x64-based Systems (KB5041792).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Pending security updates is 1 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Compliant\", \"title\": \"Domain firewall is compliant\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock\", \"finding\": \"Screen lock is no machine inactivity limit; screen saver off instead of on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-09 Cumulative Update for Windows 11 Version 23H2 for x64-based Systems (KB5047039).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Pending security updates is 1 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: teamviewer\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (macos):\n- Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"The device reports teamviewer, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall inbound policy\", \"finding\": \"Found: allow (no firewall). Required: deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nHOSTNAME: dev-dev-04\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=31, Non-Compliant=13, NotApplicable=5, Pending=7, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 90; accounts: jdoe (99999)\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | minclass=0, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | Bitdefender Endpoint Security\n- malware.service | ClamAV malware protection service | NotApplicable | not used\n- malware.signature_updates | Malware signature updates | NotApplicable | not used\n- malware.signatures | Malware signatures | NotApplicable | not used\n- malware.on_access | On-access malware protection | Compliant | Bitdefender Endpoint Security\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Non-Compliant | off for: tnguyen\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: tnguyen (3600s)\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: tnguyen\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (tnguyen)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 31 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 353 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Compliant | 4 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Timeshift\n- gdpr.backup_recent | Recent backup | Non-Compliant | 44 days ago\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Hugging Face models, Alpaca (Ollama client)\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Hugging Face models, Alpaca (Ollama client)\n- aiact.risk_classification | AI system risk classification | Pending | Hugging Face models, Alpaca (Ollama client)\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- password.complexity: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- gdpr.backup_recent: Medium\n- aiact.unapproved: Medium\n- org.breach_procedure: High\nSCORE (computed): compliant=31, assessed=44, percent=70\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | Pending security updates | High | 31 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 353 critical/high\n- gdpr.os_supported | Supported operating system version | High | Red Hat Enterprise Linux 8.10 (Ootpa)\n- org.breach_procedure | Breach response procedure | High | partial\n- password.complexity | Password complexity | Medium | minclass=0, required classes=0\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.os_supported: Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance."}, {"role": "assistant", "content": "{\"device\": \"dev-dev-04 - Red Hat Enterprise Linux 8.10 (Ootpa) 8.10\", \"score\": {\"compliant\": 31, \"assessed\": 44, \"percent\": 70}, \"counts\": {\"Compliant\": 31, \"Non-Compliant\": 13, \"NotApplicable\": 5, \"Pending\": 7, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"31 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"353 critical/high\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Red Hat Enterprise Linux 8.10 (Ootpa)\"}, {\"check_id\": \"org.breach_procedure\", \"name\": \"Breach response procedure\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"password.complexity\", \"name\": \"Password complexity\", \"risk\": \"Medium\", \"found\": \"minclass=0, required classes=0\"}], \"summary\": \"dev-dev-04 meets 31 of 44 assessed checks (70%). The most serious gaps are Pending security updates, Known vulnerabilities in installed software and Supported operating system version. 7 items are pending.\", \"next_steps\": [\"Pending security updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Supported operating system version: Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\", \"Answer the 5 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"Firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history matches the required setting\", \"finding\": \"Found: 24. This satisfies the baseline requirement of >= 24.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 313 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 832 packages checked against rhel:9 advisories. dotnet-apphost-pack-6.0 0:6.0.7-1.el9_0 (22 CVEs, critical, actively exploited; fixed in 0:6.0.35-1.el9_4; e.g. CVE-2022-38013, CVE-2022-41032); nginx-mod-mail 1:1.22.1-3.module+el9.2.0.z+20353+5a828d50.1 (14 CVEs, high, actively exploited; fixed in 1:1.24.0-7.module+el9.8.0+24379+0344c460.2; e.g. CVE-2022-41741, CVE-2022-41742); kpatch-patch-5_14_0-162_6_1-debugsource 0:1-1.el9_1 (7 CVEs, high, actively exploited; fixed in 0:1-4.el9_1; e.g. CVE-2022-3564, CVE-2022-4378); open-vm-tools-desktop 0:11.3.5-1.el9_0.1 (5 CVEs, high, actively exploited; fixed in 0:12.5.0-1.el9_6.2; e.g. CVE-2023-20867, CVE-2023-20900); samba-winbind-modules 0:4.16.4-103.el9_1 (13 CVEs, critical; fixed in 0:4.23.5-10.el9_8; e.g. CVE-2022-2127, CVE-2023-34966); and 41 more. 78 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Found: 313 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates: does not meet the baseline\", \"finding\": \"App updates is automatic app updates off instead of automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nHOSTNAME: DESKTOP-PLEBHMM\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=12, Non-Compliant=37, NotApplicable=7, Pending=6, Error=2\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 8\n- password.history | Password history | Non-Compliant | 10\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 20\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout configured\n- lockout.observation | Lockout observation period | Non-Compliant | 5 minutes\n- malware.antivirus | Antivirus | Compliant | Sophos Endpoint (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Error | not read\n- firewall.domain.outbound | Domain outbound | Non-Compliant | block\n- firewall.private.enabled | Private firewall | Non-Compliant | off\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Non-Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | block\n- screen_lock.enabled | Screen lock | Non-Compliant | no machine inactivity limit; no signed-in user to check\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.password | Screen lock password requirement | Non-Compliant | screen saver does not require sign-in for: FABRIKAM\\sali\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an enabled administrator\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (FABRIKAM\\sali)\n- encryption.feature | BitLocker | Non-Compliant | off\n- encryption.enabled | Encryption | Non-Compliant | EncryptionInProgress (42%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, None\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 1 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 152 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 23 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Enterprise build 26100\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: none; account management: none\n- gdpr.time_sync | Time synchronisation | Non-Compliant | type NoSync, service Stopped/Manual\n- gdpr.backup_configured | Backup configured | Error | not read\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Non-Compliant | enabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | chromeremotedesktop, rdp\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Hugging Face models\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Hugging Face models\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Hugging Face models\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | NotApplicable | Recall not available on this device\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- firewall.domain.outbound: Low\n- firewall.private.enabled: High\n- firewall.private.inbound: High\n- firewall.private.outbound: Low\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- user.account_type: High\n- encryption.feature: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.security_logging: Medium\n- gdpr.time_sync: Low\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.guest_account: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=12, assessed=49, percent=24\nTOP GAPS (computed, highest risk first):\n- firewall.private.enabled | Private firewall | High | off\n- firewall.private.inbound | Private inbound | High | allow\n- firewall.public.enabled | Public firewall | High | off\n- firewall.public.inbound | Public inbound | High | allow\n- user.account_type | Signed-in user account type | High | Administrator (FABRIKAM\\sali)\nFIX GUIDANCE (highest-risk gaps):\n- firewall.private.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.private.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- firewall.public.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-PLEBHMM - Microsoft Windows 11 Enterprise 10.0.26100\", \"score\": {\"compliant\": 12, \"assessed\": 49, \"percent\": 24}, \"counts\": {\"Compliant\": 12, \"Non-Compliant\": 37, \"NotApplicable\": 7, \"Pending\": 6, \"Error\": 2}, \"top_gaps\": [{\"check_id\": \"firewall.private.enabled\", \"name\": \"Private firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.public.enabled\", \"name\": \"Public firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.public.inbound\", \"name\": \"Public inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (FABRIKAM\\\\sali)\"}], \"summary\": \"DESKTOP-PLEBHMM meets 12 of 49 assessed checks (24%). The most serious gaps are Private firewall, Private inbound and Public firewall. 2 checks could not be read. 6 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Private firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Private inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Public firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Answer the 4 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"The device reports partial, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 26.7 (ios)\nAI TOOLS FOUND: ChatGPT, Microsoft Copilot\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Microsoft Copilot: Cloud AI assistant; provider Microsoft (US); data: Microsoft cloud; enterprise plans keep data under Microsoft's commercial data protection terms; default risk minimal; GDPR: Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Microsoft Copilot\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Microsoft Copilot.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Device encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (Android 10 encrypts by default)\nREQUIRED: on\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (android):\n- Update to a supported Android version (10 or later encrypts by default) or replace the phone; set a PIN or password screen lock.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Device encryption is compliant\", \"finding\": \"Found: on (Android 10 encrypts by default). This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: sali, jdoe\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (macos):\n- Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: sali, jdoe. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"Found: not running. Required: on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"The device reports no lockout configured, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised is non-compliant\", \"finding\": \"Usage & diagnostics sharing minimised is usage & diagnostics on instead of usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates is non-compliant\", \"finding\": \"The device reports freshclam not running, but the baseline requires automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; no signed-in user to check\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"Found: no machine inactivity limit; no signed-in user to check. Required: on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 15.6 (macos)\nHOSTNAME: MKT-MacBook-43\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=45, Non-Compliant=2, NotApplicable=4, Pending=3, Error=2\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5392\n- malware.service | Apple malware protection service | Error | not read\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5290, 3 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 120 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 0 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (lmuller)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Error | not read\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 25 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 15.6\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | Time Machine\n- gdpr.backup_recent | Recent backup | Compliant | 6 days ago\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | ssh, screensharing\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Claude Code\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Intel Mac\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\nSCORE (computed): compliant=45, assessed=47, percent=96\nTOP GAPS (computed, highest risk first):\n- gdpr.remote_access | Remote access services restricted | High | ssh, screensharing\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | Claude Code\nFIX GUIDANCE (highest-risk gaps):\n- gdpr.remote_access: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it."}, {"role": "assistant", "content": "{\"device\": \"MKT-MacBook-43 - macOS 15.6\", \"score\": {\"compliant\": 45, \"assessed\": 47, \"percent\": 96}, \"counts\": {\"Compliant\": 45, \"Non-Compliant\": 2, \"NotApplicable\": 4, \"Pending\": 3, \"Error\": 2}, \"top_gaps\": [{\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"ssh, screensharing\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"Claude Code\"}], \"summary\": \"MKT-MacBook-43 meets 45 of 47 assessed checks (96%). The most serious gaps are Remote access services restricted. 2 checks could not be read. 3 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Remote access services restricted: Turn off Remote Login, Screen Sharing, File Sharing and Remote Management in System Settings > General > Sharing, and uninstall unapproved tools (e.g. the Chrome Remote Desktop host).\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"Data protection impact assessment is no instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall on, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall inbound policy\", \"finding\": \"Firewall inbound policy is firewall on, stealth mode off instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 16 (android)\nAI TOOLS FOUND: ChatGPT\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Public firewall is non-compliant\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"ClamAV malware protection service is non-compliant\", \"finding\": \"ClamAV malware protection service is not running instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 15 (android)\nAI TOOLS FOUND: DeepSeek, Grok\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- DeepSeek: Cloud AI assistant; provider DeepSeek (China); data: Provider cloud in the People's Republic of China; default risk minimal; GDPR: No EU adequacy decision for China and no business data processing terms; EU data protection authorities have restricted or investigated it. Do not use with work or personal data.\n- Grok: Cloud AI assistant; provider xAI (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Consumer service; needs an enterprise agreement and a transfer basis before work or personal data is entered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"DeepSeek\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No EU adequacy decision for China and no business data processing terms; EU data protection authorities have restricted or investigated it. Do not use with work or personal data.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Grok\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer service; needs an enterprise agreement and a transfer basis before work or personal data is entered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: DeepSeek, Grok.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"The device reports no machine inactivity limit; screen saver off, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2025-11 (11 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2025-11 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Android security update\", \"finding\": \"The user reports security patch 2025-11 (11 months old), but the baseline requires <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAI TOOLS FOUND: Hugging Face models, Alpaca (Ollama client)\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /home/jdoe/.cache/huggingface/hub (4G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Hugging Face models: Local model files; provider Various model publishers; data: On the device; default risk minimal; GDPR: Record which models are present and their licences; the risk class depends on what the models are used for.\n- Alpaca (Ollama client): Desktop client for local models; provider Open source (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Hugging Face models\", \"kind\": \"Local model files\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Record which models are present and their licences; the risk class depends on what the models are used for.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Alpaca (Ollama client)\", \"kind\": \"Desktop client for local models\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Hugging Face models, Alpaca (Ollama client).\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically: does not meet the baseline\", \"finding\": \"Date & time set automatically is set by hand instead of set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › General › Date & Time: turn on Set Automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Install unknown apps blocked: does not meet the baseline\", \"finding\": \"Install unknown apps blocked is unknown apps allowed for at least one app instead of all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode on\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Firewall inbound policy is firewall off, stealth mode on instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"Compliant\", \"title\": \"Tamper protection matches the required setting\", \"finding\": \"Found: on. This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Human oversight for high-risk AI is no instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, XtsAes128\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker protection\", \"finding\": \"BitLocker protection is protection Off, XtsAes128 instead of protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 10 (android)\nAI TOOLS FOUND: Gemini, Otter\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\n- Otter: AI meeting transcription; provider Otter.ai (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Records and transcribes other people's voices: needs a lawful basis and notice to meeting participants, plus a data processing agreement.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Otter\", \"kind\": \"AI meeting transcription\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Records and transcribes other people's voices: needs a lawful basis and notice to meeting participants, plus a data processing agreement.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Gemini, Otter.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI register\", \"finding\": \"The device reports no, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] ISO 27001:2022 A.7.10 Storage media: ISO 27001:2022 A.7.10 (Storage media): Manage removable media through its life cycle according to the classification of the data on it. An auditor looks for: Rules for USB and external drives; Encryption or blocking of removable media. On endpoints: Unrestricted USB writing lets data leave the device unencrypted.\n[2] GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption): Where appropriate, protect personal data with pseudonymisation and encryption. An auditor looks for: Encrypted devices and removable media. On endpoints: Disk encryption is the standard measure for laptops holding personal data.\n[3] Encryption on Linux: Encryption on Linux. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: The system disk is fully encrypted (BitLocker / LUKS / FileVault). Phones: iOS data protection is on when a passcode is set; Android 10+ encrypts by default. How to fix on Linux: (1) Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[4] Removable media encryption on Linux: Removable media encryption on Linux. Why it matters: Personal data copied to removable media must be encrypted, or copying must be blocked. Risk if missing: Medium - A lost unencrypted USB drive with personal data is a reportable breach. Requirement: Data cannot be written to unencrypted removable drives. How to fix on Linux: (1) Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives. How to verify: Re-run the audit; Removable media encryption should be Compliant. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption); GDPR Art. 34 (Communication of a personal data breach to the data subject).\n\nMY AUDIT: Ubuntu 20.04.6 LTS 20.04 (linux), audited 2026-10-06; 41 of 49 assessed checks compliant; 8 gaps, 3 pending, 0 not read. Gaps: On-access malware protection: not running; USB storage write: allow; Pending security updates: 19 pending; Outdated third-party applications: 50 outdated; Supported operating system version: Ubuntu 20.04.6 LTS; Removable media encryption: USB storage writable without enforced encryption; Diagnostic data minimised: popularity-contest on; AI transparency to users: partial.\n\nQUESTION: How do I fix removable media encryption?"}, {"role": "assistant", "content": "Your audit found Removable media encryption: USB storage writable without enforced encryption. Personal data copied to removable media must be encrypted, or copying must be blocked. To fix it [4]:\n1. Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "linux", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-R2S2I1D\\jdoe, DESKTOP-R2S2I1D\\Administrator\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts is non-compliant\", \"finding\": \"Unauthorized employee local administrator accounts is DESKTOP-R2S2I1D\\\\jdoe, DESKTOP-R2S2I1D\\\\Administrator instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Device encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on (Android 12 encrypts by default)\nREQUIRED: on\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (android):\n- Update to a supported Android version (10 or later encrypts by default) or replace the phone; set a PIN or password screen lock.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Device encryption is compliant\", \"finding\": \"The device reports on (Android 12 encrypts by default), which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: not rooted\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Pending\", \"title\": \"Not rooted is pending\", \"finding\": \"The user could not find or confirm this setting. Current state: not answered. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Operating system AI features restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: none built in\nREQUIRED: restricted by policy\nDETAIL: Linux distributions ship no built-in AI assistant.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (linux):\n- No action needed: Linux distributions have no built-in AI assistant.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Operating system AI features restricted: not applicable\", \"finding\": \"Operating system AI features restricted is not assessed here: Linux distributions ship no built-in AI assistant.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: itadmin\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (macos):\n- Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Compliant\", \"title\": \"Approved IT administrator is compliant\", \"finding\": \"Approved IT administrator is set to itadmin, in line with the required itadmin.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (FABRIKAM\\sali)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Found: Administrator (FABRIKAM\\\\sali). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apps only from the App Store\", \"finding\": \"Apps only from the App Store is apps from outside the App Store instead of App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification: not applicable\", \"finding\": \"Not applicable on this device: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"The device reports no lockout, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Found: 5. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\", \"For a stand-alone device: net accounts /uniquepw:24\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: minclass=0, required classes=4\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Compliant\", \"title\": \"Password complexity: meets the baseline\", \"finding\": \"Found: minclass=0, required classes=4. This satisfies the baseline requirement of 3 or more character classes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Data processing agreements is no instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Windows Defender (off, out of date)\nREQUIRED: enabled and up to date\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (windows):\n- Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Antivirus: does not meet the baseline\", \"finding\": \"The device reports Windows Defender (off, out of date), but the baseline requires enabled and up to date. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 2 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Compliant\", \"title\": \"Disk encryption recovery: meets the baseline\", \"finding\": \"Found: 2 key slots, 0 unlock tokens. This satisfies the baseline requirement of recovery key or second unlock method.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 99999\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"The device reports PASS_MAX_DAYS 99999, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: none; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security event logging\", \"finding\": \"Security event logging is logon: none; account management: none instead of logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 94 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 22 applications inventoried, 11 recognised (23 common applications covered). Firefox 155.0.0 (94 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100756, CVE-2026-100767). 14 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"The device reports 94 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: screen saver does not require sign-in for: DEVSEIS\\pmartin\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"The device reports screen saver does not require sign-in for: DEVSEIS\\\\pmartin, but the baseline requires sign-in required to unlock. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"The device reports not enforced, but the baseline requires enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Found: partial. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: minclass=2, required classes=1\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Found: minclass=2, required classes=1. Required: 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (android):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off (no passcode)\nREQUIRED: on\nDETAIL: Without a passcode, iOS data protection is not active.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Data protection (encryption) is non-compliant\", \"finding\": \"Data protection (encryption) is off (no passcode) instead of on. Full-disk encryption keeps data unreadable if the device is lost or stolen. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: enforced\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Error\", \"title\": \"Password complexity: could not be checked\", \"finding\": \"This check returned an error: Value could not be read. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Compliant\", \"title\": \"Domain inbound: meets the baseline\", \"finding\": \"The device reports block, which meets the requirement (block).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 58 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 58\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"The device reports 58 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"Found: not set. Required: >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with pinHistory = 24.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 14 (android)\nAI TOOLS FOUND: Character.AI, ChatGPT, Gemini\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Character.AI: AI companion chatbot; provider Character Technologies (US); data: Provider cloud outside the EEA; default risk limited; GDPR: Consumer entertainment chatbot; not suitable for work use. Users must be told they are talking to an AI (AI Act Art. 50).\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Character.AI\", \"kind\": \"AI companion chatbot\", \"approved\": false, \"default_risk_class\": \"limited\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer entertainment chatbot; not suitable for work use. Users must be told they are talking to an AI (AI Act Art. 50).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Character.AI, ChatGPT, Gemini.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold is non-compliant\", \"finding\": \"Found: 10. Required: 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\", \"Set deny = 5 in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Public outbound\", \"finding\": \"Public outbound is block instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nHOSTNAME: ops-dev-59\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=33, Non-Compliant=16, NotApplicable=4, Pending=2, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=0, required classes=4\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Non-Compliant | no signature database\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | pmartin, mkhan\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (akaya)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 46 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 179 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Compliant | 26 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.security_logging | Security event logging | Non-Compliant | auditd not running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Timeshift\n- gdpr.backup_recent | Recent backup | Compliant | 7 days ago\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Non-Compliant | LightDM guest session allowed\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Error | not read\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.service: High\n- malware.signature_updates: Medium\n- malware.signatures: High\n- malware.on_access: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- gdpr.security_logging: Medium\n- gdpr.guest_account: Medium\n- org.ropa: High\n- org.ai_transparency: Low\nSCORE (computed): compliant=33, assessed=49, percent=67\nTOP GAPS (computed, highest risk first):\n- malware.service | ClamAV malware protection service | High | not running\n- malware.signatures | Malware signatures | High | no signature database\n- admin.unauthorized | Unauthorized employee administrator accounts | High | pmartin, mkhan\n- user.account_type | Signed-in user account type | High | Administrator (akaya)\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 179 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d."}, {"role": "assistant", "content": "{\"device\": \"ops-dev-59 - Red Hat Enterprise Linux 8.10 (Ootpa) 8.10\", \"score\": {\"compliant\": 33, \"assessed\": 49, \"percent\": 67}, \"counts\": {\"Compliant\": 33, \"Non-Compliant\": 16, \"NotApplicable\": 4, \"Pending\": 2, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"no signature database\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"pmartin, mkhan\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (akaya)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"179 critical/high\"}], \"summary\": \"ops-dev-59 meets 33 of 49 assessed checks (67%). The most serious gaps are ClamAV malware protection service, Malware signatures and Unauthorized employee administrator accounts. 1 checks could not be read. 2 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Compliant\", \"title\": \"Find My Device: meets the baseline\", \"finding\": \"Found: on. This satisfies the baseline requirement of on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-J3GFH4F\\mkhan, DESKTOP-J3GFH4F\\Administrator\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts is non-compliant\", \"finding\": \"Found: DESKTOP-J3GFH4F\\\\mkhan, DESKTOP-J3GFH4F\\\\Administrator. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude, Microsoft Copilot\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: Claude, Microsoft Copilot. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 8\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Found: 8. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Compliant\", \"title\": \"AI transparency to users: meets the baseline\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 50\"]}"}], "meta": {"task": "finding", "os": "android", "audit": 7000367, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 99999\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"The device reports PASS_MAX_DAYS 99999, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: App updates\", \"finding\": \"App updates is automatic app updates off instead of automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 12 (android)\nAI TOOLS FOUND: Microsoft Copilot\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Microsoft Copilot: Cloud AI assistant; provider Microsoft (US); data: Microsoft cloud; enterprise plans keep data under Microsoft's commercial data protection terms; default risk minimal; GDPR: Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Microsoft Copilot\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Microsoft Copilot.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"The device reports not set, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"Found: 0 days. Required: >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold: does not meet the baseline\", \"finding\": \"The device reports 10, but the baseline requires 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Screen timeout 30 minutes\nREQUIRED: <= 300 seconds\nDETAIL: The phone stays unlocked for 30 minutes when not in use.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen timeout: does not meet the baseline\", \"finding\": \"The user reports Screen timeout 30 minutes, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: AUOptions=2\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates: does not meet the baseline\", \"finding\": \"The device reports AUOptions=2, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001464, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 15.7.2 (macos)\nAI TOOLS FOUND: LM Studio, Ollama\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /Users/jdoe/.ollama/models (51G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- LM Studio: Local model runner; provider Element Labs (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences and whether its local server is exposed.\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"LM Studio\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences and whether its local server is exposed.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: LM Studio, Ollama.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI register\", \"finding\": \"The device reports partial, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"Found: none running. Required: antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities: does not meet the baseline\", \"finding\": \"Record of processing activities is partial instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates is non-compliant\", \"finding\": \"Automatic system updates is manual instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates is non-compliant\", \"finding\": \"The user reports automatic app updates off, but the baseline requires automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Found: not configured. Required: required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound is non-compliant\", \"finding\": \"The device reports block, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nDETAIL: pam_pwquality not enabled; pam_unix minimum\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Found: 6. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: HR-MacBook-58\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=30, Non-Compliant=18, NotApplicable=5, Pending=2, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 10\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5282\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5308, 3 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall on, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 600 idle, password delay 60\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 0 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | ewagner, tnguyen\n- user.account_type | Signed-in user account type | Compliant | Standard (rschmit)\n- encryption.enabled | Encryption | Non-Compliant | FileVault off\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no FileVault\n- vuln.pending_updates | Pending security updates | Error | not read\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 53 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Compliant | 12 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | ArqAgent\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | restricted by profile\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- lockout.duration: Low\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.removable_encryption: Medium\n- org.ropa: High\n- org.dpia: Medium\n- org.ai_register: Medium\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=30, assessed=48, percent=63\nTOP GAPS (computed, highest risk first):\n- firewall.inbound | Firewall inbound policy | High | firewall on, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | ewagner, tnguyen\n- encryption.enabled | Encryption | High | FileVault off\n- encryption.protection | Disk encryption protection | High | not encrypted\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 53 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\n- encryption.enabled: Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM."}, {"role": "assistant", "content": "{\"device\": \"HR-MacBook-58 - macOS 26.0.1\", \"score\": {\"compliant\": 30, \"assessed\": 48, \"percent\": 63}, \"counts\": {\"Compliant\": 30, \"Non-Compliant\": 18, \"NotApplicable\": 5, \"Pending\": 2, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall on, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"ewagner, tnguyen\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"FileVault off\"}, {\"check_id\": \"encryption.protection\", \"name\": \"Disk encryption protection\", \"risk\": \"High\", \"found\": \"not encrypted\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"53 critical/high\"}], \"summary\": \"HR-MacBook-58 meets 30 of 48 assessed checks (63%). The most serious gaps are Firewall inbound policy, Unauthorized employee administrator accounts and Encryption. 1 checks could not be read. 2 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Encryption: Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"The device reports 0, but the baseline requires >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Jan, Hugging Face models\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Found: Jan, Hugging Face models. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Analytics sharing minimised\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: OPS-MacBook-22\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=29, Non-Compliant=18, NotApplicable=6, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | not set\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5315\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Non-Compliant | XProtect 5272, 61 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall on, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 600 idle, password delay 5\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Compliant | password after 5 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (pmartin)\n- encryption.enabled | Encryption | Non-Compliant | FileVault off\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | no FileVault\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 39 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Compliant | 27 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | external drives writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | remotemanagement\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | restricted by profile\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- malware.signatures: High\n- firewall.inbound: High\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- user.account_type: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.ropa: High\n- org.dpia: Medium\n- org.processor_agreements: High\nSCORE (computed): compliant=29, assessed=47, percent=62\nTOP GAPS (computed, highest risk first):\n- malware.signatures | Malware protection signatures | High | XProtect 5272, 61 days old\n- firewall.inbound | Firewall inbound policy | High | firewall on, stealth mode off\n- user.account_type | Signed-in user account type | High | Administrator (pmartin)\n- encryption.enabled | Encryption | High | FileVault off\n- encryption.protection | Disk encryption protection | High | not encrypted\nFIX GUIDANCE (highest-risk gaps):\n- malware.signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- user.account_type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account."}, {"role": "assistant", "content": "{\"device\": \"OPS-MacBook-22 - macOS 26.0.1\", \"score\": {\"compliant\": 29, \"assessed\": 47, \"percent\": 62}, \"counts\": {\"Compliant\": 29, \"Non-Compliant\": 18, \"NotApplicable\": 6, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.signatures\", \"name\": \"Malware protection signatures\", \"risk\": \"High\", \"found\": \"XProtect 5272, 61 days old\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall on, stealth mode off\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (pmartin)\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"FileVault off\"}, {\"check_id\": \"encryption.protection\", \"name\": \"Disk encryption protection\", \"risk\": \"High\", \"found\": \"not encrypted\"}], \"summary\": \"OPS-MacBook-22 meets 29 of 47 assessed checks (62%). The most serious gaps are Malware protection signatures, Firewall inbound policy and Signed-in user account type. 3 items are pending.\", \"next_steps\": [\"Malware protection signatures: Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Signed-in user account type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\", \"Answer the 3 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Public firewall is non-compliant\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates: does not meet the baseline\", \"finding\": \"Found: automatic app updates off. Required: automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: firewalld active\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Compliant\", \"title\": \"Firewall is compliant\", \"finding\": \"Firewall is set to firewalld active, in line with the required active.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: popularity-contest on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"The device reports popularity-contest on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Tamper protection\", \"finding\": \"The device reports off, but the baseline requires on. Tamper protection stops malware or users switching off Defender.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Attackers commonly disable antivirus before deploying ransomware.\"}, \"recommendation\": [\"Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Tamper protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (windows):\n- Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > System > Removable Storage Access > Removable Disks: Deny write access = Enabled, or require BitLocker To Go ('Deny write access to removable drives not protected by BitLocker').\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Compliant\", \"title\": \"Firewall inbound policy matches the required setting\", \"finding\": \"The device reports deny, which meets the requirement (deny).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Screen timeout 2 minutes\nREQUIRED: <= 300 seconds\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen timeout: meets the baseline\", \"finding\": \"Screen timeout is set to Screen timeout 2 minutes, in line with the required <= 300 seconds. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: minclass=0, required classes=0\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Password complexity is minclass=0, required classes=0 instead of 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 99999\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"Found: PASS_MAX_DAYS 99999. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"Found: off. Required: on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound: does not meet the baseline\", \"finding\": \"Found: block. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Screen timeout 30 minutes\nREQUIRED: <= 300 seconds\nDETAIL: The phone stays unlocked for 30 minutes when not in use.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen timeout is non-compliant\", \"finding\": \"The user reports Screen timeout 30 minutes, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates: does not meet the baseline\", \"finding\": \"The user reports manual, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 14.7.8 (macos)\nAI TOOLS FOUND: LM Studio\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- LM Studio: Local model runner; provider Element Labs (runs locally); data: On the device; default risk minimal; GDPR: No transfer to a provider; check model licences and whether its local server is exposed.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"LM Studio\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider; check model licences and whether its local server is exposed.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: LM Studio.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Data protection impact assessment is no instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode on\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Firewall inbound policy is firewall off, stealth mode on instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Restrict external storage with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read: meets the baseline\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 40 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 39; snaps: firefox\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Outdated third-party applications is 40 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Compliant\", \"title\": \"Public firewall matches the required setting\", \"finding\": \"Found: on. This satisfies the baseline requirement of on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Supported operating system version on Linux: Supported operating system version on Linux. Why it matters: An operating system that no longer receives security updates cannot be kept secure. Risk if missing: High - New vulnerabilities in an unsupported OS are never fixed. Requirement: The OS version still receives security updates (minimums in the baseline). Phones: IOS_MIN_MAJOR / ANDROID_MIN_MAJOR. How to fix on Linux: (1) Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance. How to verify: Re-run the audit; Supported operating system version should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Backup configured on Linux: Backup configured on Linux. Why it matters: Backups make it possible to restore personal data after loss, theft, failure or ransomware. Risk if missing: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware. Requirement: A backup solution is configured for user data. How to fix on Linux: (1) Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup). How to verify: Re-run the audit; Backup configured should be Compliant. Related controls: ISO 27001:2022 A.8.13 (Information backup); GDPR Art. 32(1)(c) (Security of processing: restore availability and access after an incident).\n[3] Unapproved AI tools (shadow AI) on Linux: Unapproved AI tools (shadow AI) on Linux. Why it matters: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Risk if missing: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations. Requirement: Every AI tool found is listed in APPROVED_AI_TOOLS. How to fix on Linux: (1) Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it. (2) Publish an AI acceptable-use policy so staff know which tools are allowed. (3) Remove unapproved tools with the package manager and restrict installation rights. How to verify: Re-run the audit; Unapproved AI tools should read none. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 28 (Processor); GDPR Art. 44 (General principle for transfers); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[4] Operating system AI features restricted on Linux: Operating system AI features restricted on Linux. Why it matters: Built-in OS AI features can capture screen content or send text to external AI services. Risk if missing: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose. Requirement: OS-level AI features that capture or send user data are disabled by policy. NotApplicable where the OS has none. Phones: external AI integrations (ChatGPT in Apple Intelligence, assistant access to screen content) are off unless approved. How to fix on Linux: (1) No action needed: Linux distributions have no built-in AI assistant. How to verify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer (x86-64), Chrome 152; detected by the browser, security settings not checked.\n\nQUESTION: How can I set up operating system ai features restricted on my Linux?"}, {"role": "assistant", "content": "Built-in OS AI features can capture screen content or send text to external AI services. To fix it on Linux [4]:\n1. No action needed: Linux distributions have no built-in AI assistant.\n\nTo check it worked: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}], "meta": {"task": "chat", "os": "linux", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period matches the required setting\", \"finding\": \"Found: 30 minutes. This satisfies the baseline requirement of >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: ChatGPT. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: <= 0 pending\nDETAIL: Apple's update service could not be reached.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Error\", \"title\": \"Pending security updates: evidence missing\", \"finding\": \"This check returned an error: Apple's update service could not be reached. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Android 11\nREQUIRED: >= Android 13\nDETAIL: Version reported by the browser (Chrome 138). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (android):\n- Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"The device reports Android 11, but the baseline requires >= Android 13. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answer ORG_DPIA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; no signed-in user to check\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Screen lock timeout is no machine inactivity limit; no signed-in user to check instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Usage & diagnostics sharing minimised\", \"finding\": \"The user reports usage & diagnostics on, but the baseline requires usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: automatic\nREQUIRED: automatic\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic system updates is compliant\", \"finding\": \"The user reports automatic, which meets the requirement (automatic). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Compliant\", \"title\": \"Data processing agreements: meets the baseline\", \"finding\": \"Data processing agreements is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Password complexity is not enforced instead of enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"Password history is 5 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold is non-compliant\", \"finding\": \"Found: 10. Required: 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\", \"Set deny = 5 in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 134 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-05-21 22:24:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently is non-compliant\", \"finding\": \"Security updates installed recently is 134 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nHOSTNAME: ops-ws-59\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=27, Non-Compliant=21, NotApplicable=4, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 12\n- password.history | Password history | Non-Compliant | 12\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Non-Compliant | minclass=0, required classes=0\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Non-Compliant | 66 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Non-Compliant | deny\n- usb.read | USB storage read | Non-Compliant | deny\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | mkhan\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (lmuller)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 3 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 4 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 26.04.1 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | restic-backup.timer\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- malware.antivirus: High\n- malware.service: High\n- malware.signatures: High\n- malware.on_access: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- usb.device_access: Low\n- usb.read: Low\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- gdpr.telemetry: Low\n- org.breach_procedure: High\n- org.ai_register: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=27, assessed=48, percent=56\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.service | ClamAV malware protection service | High | not running\n- malware.signatures | Malware signatures | High | 66 days old\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now."}, {"role": "assistant", "content": "{\"device\": \"ops-ws-59 - Ubuntu 26.04.1 LTS 26.04\", \"score\": {\"compliant\": 27, \"assessed\": 48, \"percent\": 56}, \"counts\": {\"Compliant\": 27, \"Non-Compliant\": 21, \"NotApplicable\": 4, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"66 days old\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}], \"summary\": \"ops-ws-59 meets 27 of 48 assessed checks (56%). The most serious gaps are Antivirus, ClamAV malware protection service and Malware signatures. 4 items are pending.\", \"next_steps\": [\"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Answer the 3 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: guest login off\nREQUIRED: guest login off\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (macos):\n- sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled is compliant\", \"finding\": \"The device reports guest login off, which meets the requirement (guest login off).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Operating system AI features restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: none built in\nREQUIRED: restricted by policy\nDETAIL: Linux distributions ship no built-in AI assistant.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (linux):\n- No action needed: Linux distributions have no built-in AI assistant.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Operating system AI features restricted: not assessed on this device\", \"finding\": \"This check does not apply: Linux distributions ship no built-in AI assistant.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google Play Protect\", \"finding\": \"Google Play Protect is Play Protect scanning off instead of on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 17 applications inventoried, 10 recognised (23 common applications covered). zoom.us 6.5.5 (6 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-62483, CVE-2025-64739). 3 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 6 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Supported operating system version on Linux: Supported operating system version on Linux. Why it matters: An operating system that no longer receives security updates cannot be kept secure. Risk if missing: High - New vulnerabilities in an unsupported OS are never fixed. Requirement: The OS version still receives security updates (minimums in the baseline). Phones: IOS_MIN_MAJOR / ANDROID_MIN_MAJOR. How to fix on Linux: (1) Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance. How to verify: Re-run the audit; Supported operating system version should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Operating system AI features restricted on Linux: Operating system AI features restricted on Linux. Why it matters: Built-in OS AI features can capture screen content or send text to external AI services. Risk if missing: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose. Requirement: OS-level AI features that capture or send user data are disabled by policy. NotApplicable where the OS has none. Phones: external AI integrations (ChatGPT in Apple Intelligence, assistant access to screen content) are off unless approved. How to fix on Linux: (1) No action needed: Linux distributions have no built-in AI assistant. How to verify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[3] ISO 27001:2022 A.8.8 Management of technical vulnerabilities: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities): Find technical weaknesses in time, judge the exposure and fix them. An auditor looks for: Timely patching; Supported operating system versions. On endpoints: Automatic updates, recent patch dates and a supported OS show vulnerabilities are being managed.\n[4] How to run an audit of a computer: A web page cannot read a computer's settings, so computers are audited with the free Devseis desktop app for Mac, Windows (beta) and Linux. Download it from the Download page, install it, click Run audit and enter the computer password once. The first run downloads the AI auditor; the audit then takes about 5 to 15 minutes and ends with a PDF report.\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer (x86-64), Chrome 152; detected by the browser, security settings not checked.\n\nQUESTION: What version of my operating system do I have?"}, {"role": "assistant", "content": "This browser reports Linux on a Linux computer (x86-64), using Chrome 152. That is all a web page can see: it can't read security settings such as encryption, the firewall or the screen lock. To check those, run an audit with the free Devseis desktop app [4]."}], "meta": {"task": "chat", "os": "linux", "status": "device", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Compliant\", \"title\": \"AI register is compliant\", \"finding\": \"AI register is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall outbound policy is non-compliant\", \"finding\": \"Firewall outbound policy is no firewall instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not covered\nREQUIRED: <= 0 critical/high\nDETAIL: Ubuntu 20.04.6 LTS is not in the vulnerability data (covered: ubuntu:22.04, ubuntu:24.04, ubuntu:25.10, ubuntu:26.04, ubuntu:23.10, ubuntu:24.10, ubuntu:25.04, debian:12, debian:13, debian:14, debian:11, rhel:8, rhel:9, rhel:10, almalinux:8, almalinux:9, almalinux:10, rocky:8, rocky:9, rocky:10).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Pending\", \"title\": \"Known vulnerabilities in installed software: decision pending\", \"finding\": \"Ubuntu 20.04.6 LTS is not in the vulnerability data (covered: ubuntu:22.04, ubuntu:24.04, ubuntu:25.10, ubuntu:26.04, ubuntu:23.10, ubuntu:24.10, ubuntu:25.04, debian:12, debian:13, debian:14, debian:11, rhel:8, rhel:9, rhel:10, almalinux:8, almalinux:9, almalinux:10, rocky:8, rocky:9, rocky:10). Current state: not covered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 37 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: libxml2, vim-minimal, glibc-common. Uses the cached repository data.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Pending security updates is 37 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: asmith (1800s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout is non-compliant\", \"finding\": \"Found: too long for: asmith (1800s). Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Minimum password length is 6 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: within limit\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: rschmit\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout matches the required setting\", \"finding\": \"Screen lock timeout is set to within limit, in line with the required <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock is non-compliant\", \"finding\": \"The user reports Auto-Lock Never, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"Found: partial. Required: yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Compliant\", \"title\": \"Data processing agreements is compliant\", \"finding\": \"Data processing agreements is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] What is in the audit report: The report has a compliance score, the most important gaps, and for each problem: what was found, why it matters, the risk level, step-by-step fixes for that operating system, how to verify the fix, and the related ISO 27001, GDPR and EU AI Act references. It also lists AI tools found and their EU AI Act risk class.\n[2] Privacy: where do audit results go?: Nowhere. Evidence and reports stay on the device. The app downloads only two things: the AI auditor model and a signed list of known vulnerabilities. Nothing about the device is uploaded. Chat history is stored only in this browser.\n[3] Automatic security updates on Windows: Automatic security updates on Windows. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on Windows: (1) Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled. How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Minimum password length on Windows: Minimum password length on Windows. Why it matters: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength. Risk if missing: Medium - Weak passwords make account takeover through guessing or credential stuffing easier. Requirement: Configured minimum length >= PASSWORD_MIN_LENGTH. How to fix on Windows: (1) Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent). (2) For a stand-alone device: net accounts /minpwlen:14 How to verify: Re-run the audit; Minimum password length should read 14 or more. Related controls: ISO 27001:2022 A.5.17 (Authentication information); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Microsoft Windows 11 Pro 10.0.22621 (windows), audited 2026-10-06; 43 of 59 assessed checks compliant; 16 gaps, 3 pending, 1 not read. Gaps: Minimum password length: 6; Password history: 0; Minimum password age: 0 days; Password complexity: disabled; Private firewall: off; Private inbound: allow; Unauthorized employee local administrator accounts: DESKTOP-WMSTQZV\\akaya, DESKTOP-WMSTQZV\\asmith; Pending security updates: 2 pending.\n\nQUESTION: What is the most important thing to fix?"}, {"role": "assistant", "content": "Your audit found 16 gaps. Start with Minimum password length (found: 6). Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength. To fix it [4]:\n1. Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n2. For a stand-alone device: net accounts /minpwlen:14\n\nThen look at: Password history, Minimum password age, Password complexity."}], "meta": {"task": "chat", "os": "windows", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Found: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 45 outdated\nREQUIRED: <= 0 outdated\nDETAIL: python@3.13 (8.1.5) < 8.1.10; wget (1.4.1) < 1.6.15; xz (7.2.0) < 7.3.15; curl (19.12.9) < 19.14.14; pkgconf (29.17.3) < 29.19.10.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (macos):\n- Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Outdated third-party applications is 45 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Malware signature updates\", \"finding\": \"The device reports freshclam not running, but the baseline requires automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Compliant\", \"title\": \"Breach response procedure is compliant\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked: does not meet the baseline\", \"finding\": \"Notification previews hidden when locked is Always instead of When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: lmuller, asmith\nREQUIRED: password required\nDETAIL: Users checked: lmuller, asmith\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"The device reports not required for: lmuller, asmith, but the baseline requires password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"Found: none. This satisfies the baseline requirement of only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"Found: assistant screen access on. Required: screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: all removable storage denied by policy\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (windows):\n- Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage read\", \"finding\": \"Found: deny. Required: allow. The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage read should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: minclass=0, required classes=0\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Found: minclass=0, required classes=0. Required: 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Recall feature Enabled, not disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Windows Recall / Copilot data analysis disabled\", \"finding\": \"The device reports Recall feature Enabled, not disabled by policy, but the baseline requires disabled by policy. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"The device reports no lockout, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LAPTOP-Y6NRCJU\\lmuller, LAPTOP-Y6NRCJU\\tnguyen\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts: does not meet the baseline\", \"finding\": \"The device reports LAPTOP-Y6NRCJU\\\\lmuller, LAPTOP-Y6NRCJU\\\\tnguyen, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 32 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: openssl, curl, openssh-server. Package lists are 14 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"The device reports 32 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nHOSTNAME: dev-lt-15\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=29, Non-Compliant=18, NotApplicable=5, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Non-Compliant | 0\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | pam_pwquality not enabled\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 0 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | firewalld active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | NotApplicable | no GNOME desktop\n- screen_lock.timeout | Screen lock timeout | NotApplicable | no GNOME desktop\n- screen_lock.password | Screen lock password requirement | NotApplicable | no GNOME desktop\n- usb.device_access | USB storage device access | Non-Compliant | deny\n- usb.read | USB storage read | Non-Compliant | deny\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (lmuller)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 37 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | unattended-upgrades installed=1, enabled=0\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 69 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 26.04.1 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Timeshift\n- gdpr.backup_recent | Recent backup | Compliant | 0 days ago\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | popularity-contest on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Claude Code\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- password.complexity: Medium\n- lockout.duration: Low\n- malware.service: High\n- malware.on_access: Medium\n- usb.device_access: Low\n- usb.read: Low\n- user.account_type: High\n- vuln.outdated_apps: Medium\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.breach_procedure: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=29, assessed=47, percent=62\nTOP GAPS (computed, highest risk first):\n- malware.service | ClamAV malware protection service | High | not running\n- user.account_type | Signed-in user account type | High | Administrator (lmuller)\n- gdpr.auto_updates | Automatic security updates | High | unattended-upgrades installed=1, enabled=0\n- gdpr.updates_recent | Security updates installed recently | High | 69 days ago\n- org.breach_procedure | Breach response procedure | High | no\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- user.account_type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\n- gdpr.auto_updates: Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer."}, {"role": "assistant", "content": "{\"device\": \"dev-lt-15 - Ubuntu 26.04.1 LTS 26.04\", \"score\": {\"compliant\": 29, \"assessed\": 47, \"percent\": 62}, \"counts\": {\"Compliant\": 29, \"Non-Compliant\": 18, \"NotApplicable\": 5, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (lmuller)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic security updates\", \"risk\": \"High\", \"found\": \"unattended-upgrades installed=1, enabled=0\"}, {\"check_id\": \"gdpr.updates_recent\", \"name\": \"Security updates installed recently\", \"risk\": \"High\", \"found\": \"69 days ago\"}, {\"check_id\": \"org.breach_procedure\", \"name\": \"Breach response procedure\", \"risk\": \"High\", \"found\": \"no\"}], \"summary\": \"dev-lt-15 meets 29 of 47 assessed checks (62%). The most serious gaps are ClamAV malware protection service, Signed-in user account type and Automatic security updates. 4 items are pending.\", \"next_steps\": [\"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Signed-in user account type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\", \"Automatic security updates: Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 76 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 22 applications inventoried, 11 recognised (31 common applications covered). Mozilla Firefox (x64 en-US) 153.3.0 (57 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100756, CVE-2026-100767); Adobe Acrobat Reader (64-bit) 26.001.21662 (19 CVEs, high; fixed in 26.002.21901; e.g. CVE-2026-79907, CVE-2026-79908). 22 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 76 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Sensitive notifications hidden when locked: does not meet the baseline\", \"finding\": \"Sensitive notifications hidden when locked is all content shown instead of sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: iOS 26.7\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Safari 26.7).\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version: meets the baseline\", \"finding\": \"Supported operating system version is set to iOS 26.7, in line with the required >= iOS 18.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: usage & diagnostics off\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Pending\", \"title\": \"Usage & diagnostics sharing minimised is pending\", \"finding\": \"The user could not find or confirm this setting. Current state: not answered. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Removable media encryption is USB drives writable without encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-09 Cumulative Update for Windows 11 Version 24H2 for x64-based Systems (KB5043984).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates: does not meet the baseline\", \"finding\": \"Pending security updates is 1 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iPadOS 18.7.9 (update to 18.7.10 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iPadOS 18.7.10 or later in Settings › General › Software Update.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"iOS security updates: does not meet the baseline\", \"finding\": \"iOS security updates is iPadOS 18.7.9 (update to 18.7.10 available) instead of latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates: does not meet the baseline\", \"finding\": \"Automatic system updates is manual instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Found: USB storage writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Restrict external storage with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Compliant\", \"title\": \"USB storage read matches the required setting\", \"finding\": \"USB storage read is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nHOSTNAME: legal-ws-59\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=41, Non-Compliant=8, NotApplicable=4, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=4, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 1 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | nftables active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (pmartin)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 3 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 19 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 50 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Pending | not covered\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 29 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Ubuntu 20.04.6 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Deja Dup (jdoe)\n- gdpr.backup_recent | Recent backup | Compliant | 2 days ago\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | popularity-contest on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- malware.on_access: Medium\n- usb.write: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- gdpr.os_supported: High\n- gdpr.removable_encryption: Medium\n- gdpr.telemetry: Low\n- org.ai_transparency: Low\nSCORE (computed): compliant=41, assessed=49, percent=84\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | Pending security updates | High | 19 pending\n- gdpr.os_supported | Supported operating system version | High | Ubuntu 20.04.6 LTS\n- malware.on_access | On-access malware protection | Medium | not running\n- usb.write | USB storage write | Medium | allow\n- vuln.outdated_apps | Outdated third-party applications | Medium | 50 outdated\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\n- gdpr.os_supported: Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\n- malware.on_access: Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning."}, {"role": "assistant", "content": "{\"device\": \"legal-ws-59 - Ubuntu 20.04.6 LTS 20.04\", \"score\": {\"compliant\": 41, \"assessed\": 49, \"percent\": 84}, \"counts\": {\"Compliant\": 41, \"Non-Compliant\": 8, \"NotApplicable\": 4, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"19 pending\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Ubuntu 20.04.6 LTS\"}, {\"check_id\": \"malware.on_access\", \"name\": \"On-access malware protection\", \"risk\": \"Medium\", \"found\": \"not running\"}, {\"check_id\": \"usb.write\", \"name\": \"USB storage write\", \"risk\": \"Medium\", \"found\": \"allow\"}, {\"check_id\": \"vuln.outdated_apps\", \"name\": \"Outdated third-party applications\", \"risk\": \"Medium\", \"found\": \"50 outdated\"}], \"summary\": \"legal-ws-59 meets 41 of 49 assessed checks (84%). The most serious gaps are Pending security updates and Supported operating system version. 3 items are pending.\", \"next_steps\": [\"Pending security updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\", \"Supported operating system version: Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\", \"On-access malware protection: Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Removable media encryption is USB drives writable without encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google Play Protect\", \"finding\": \"The user reports Play Protect scanning off, but the baseline requires on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"AI literacy training is no instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 13 (android)\nHOSTNAME: sales-pixel-05\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=13, Non-Compliant=5, NotApplicable=2, Pending=10, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 1 minute\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | pattern\n- mobile.jailbreak_root | Not rooted | Non-Compliant | rooted or bootloader unlocked\n- mobile.app_sources | Install unknown apps blocked | Pending | not answered\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 13 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-09 (1 month old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Compliant | Android 13\n- gdpr.time_sync | Date & time set automatically | Pending | not answered\n- gdpr.backup_configured | Google backup | Compliant | phone maker cloud backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- mobile.find_my_device: Medium\n- gdpr.telemetry: Low\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=13, assessed=18, percent=72\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not rooted | High | rooted or bootloader unlocked\n- mobile.passcode_strength | Screen lock strength | Medium | pattern\n- mobile.find_my_device | Find My Device | Medium | off\n- aiact.os_ai_features | AI assistant access restricted | Medium | assistant screen access on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Low | usage & diagnostics on\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\n- mobile.passcode_strength: Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\n- mobile.find_my_device: Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on."}, {"role": "assistant", "content": "{\"device\": \"sales-pixel-05 - Android 13\", \"score\": {\"compliant\": 13, \"assessed\": 18, \"percent\": 72}, \"counts\": {\"Compliant\": 13, \"Non-Compliant\": 5, \"NotApplicable\": 2, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not rooted\", \"risk\": \"High\", \"found\": \"rooted or bootloader unlocked\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Screen lock strength\", \"risk\": \"Medium\", \"found\": \"pattern\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My Device\", \"risk\": \"Medium\", \"found\": \"off\"}, {\"check_id\": \"aiact.os_ai_features\", \"name\": \"AI assistant access restricted\", \"risk\": \"Medium\", \"found\": \"assistant screen access on\"}, {\"check_id\": \"gdpr.telemetry\", \"name\": \"Usage & diagnostics sharing minimised\", \"risk\": \"Low\", \"found\": \"usage & diagnostics on\"}], \"summary\": \"sales-pixel-05 meets 13 of 18 assessed checks (72%). The most serious gaps are Not rooted. 10 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not rooted: Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\", \"Screen lock strength: Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\", \"Find My Device: Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2025-02 (20 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2025-02 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Android security update is non-compliant\", \"finding\": \"The user reports security patch 2025-02 (20 months old), but the baseline requires <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 365 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration is compliant\", \"finding\": \"Found: 365 days. This satisfies the baseline requirement of 1-365 days.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound: does not meet the baseline\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 15 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period matches the required setting\", \"finding\": \"Lockout observation period is set to 15 minutes, in line with the required >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates is non-compliant\", \"finding\": \"The user reports automatic app updates off, but the baseline requires automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (macos):\n- sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nHOSTNAME: ops-ws-08\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=11, Non-Compliant=33, NotApplicable=2, Pending=5, Error=5\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 8\n- password.history | Password history | Non-Compliant | 10\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=4, required classes=0\n- lockout.threshold | Account lockout threshold | Non-Compliant | pam_faillock not enabled\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Non-Compliant | no signature database\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Error | not read\n- firewall.inbound | Firewall inbound policy | Error | not read\n- firewall.outbound | Firewall outbound policy | Error | not read\n- screen_lock.enabled | Screen lock | Non-Compliant | off for: sali, asmith\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: sali (0s)\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: sali, asmith\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | ewagner\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (sali)\n- encryption.enabled | Encryption | Non-Compliant | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | Error | not read\n- encryption.recovery | Disk encryption recovery | Error | not read\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 59 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 313 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | dnf-automatic timer off, apply_updates=no\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 108 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Red Hat Enterprise Linux 9.6 (Plow)\n- gdpr.security_logging | Security event logging | Non-Compliant | auditd not running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | anydesk\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Claude Code\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.history: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.antivirus: High\n- malware.service: High\n- malware.signature_updates: Medium\n- malware.signatures: High\n- malware.on_access: Medium\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.enabled: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=11, assessed=44, percent=25\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.service | ClamAV malware protection service | High | not running\n- malware.signatures | Malware signatures | High | no signature database\n- admin.unauthorized | Unauthorized employee administrator accounts | High | ewagner\n- user.account_type | Signed-in user account type | High | Administrator (sali)\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now."}, {"role": "assistant", "content": "{\"device\": \"ops-ws-08 - Red Hat Enterprise Linux 9.6 (Plow) 9.6\", \"score\": {\"compliant\": 11, \"assessed\": 44, \"percent\": 25}, \"counts\": {\"Compliant\": 11, \"Non-Compliant\": 33, \"NotApplicable\": 2, \"Pending\": 5, \"Error\": 5}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"no signature database\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"ewagner\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (sali)\"}], \"summary\": \"ops-ws-08 meets 11 of 44 assessed checks (25%). The most serious gaps are Antivirus, ClamAV malware protection service and Malware signatures. 5 checks could not be read because the audit was not run as administrator. 5 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Answer the 3 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted: does not meet the baseline\", \"finding\": \"Found: ChatGPT extension on. Required: external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: usb-storage module disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Non-Compliant\", \"title\": \"USB storage device access: does not meet the baseline\", \"finding\": \"Found: deny. Required: allow. Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The device access setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage device access should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: no guest login\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled matches the required setting\", \"finding\": \"Found: no guest login. This satisfies the baseline requirement of no guest login.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: dnf-automatic timer on, apply_updates=yes\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates: meets the baseline\", \"finding\": \"Automatic security updates is set to dnf-automatic timer on, apply_updates=yes, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history: does not meet the baseline\", \"finding\": \"Password history is 5 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\", \"For a stand-alone device: net accounts /uniquepw:24\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: unknown apps blocked for all apps\nREQUIRED: all apps: Not allowed\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Compliant\", \"title\": \"Install unknown apps blocked is compliant\", \"finding\": \"Found: unknown apps blocked for all apps. This satisfies the baseline requirement of all apps: Not allowed. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamonacc running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Compliant\", \"title\": \"On-access malware protection is compliant\", \"finding\": \"Found: clamonacc running. This satisfies the baseline requirement of on-access scanning.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: winrm, chromeremotedesktop\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted is non-compliant\", \"finding\": \"The device reports winrm, chromeremotedesktop, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"The device reports USB drives writable without encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Found: share Mac analytics=1, share with app developers=0. Required: analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Malware signature updates\", \"finding\": \"Found: freshclam not running. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data protection impact assessment\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"Found: partial. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: File History (jdoe).\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Found: date not readable. Confirm the last backup date in: File History (jdoe).\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: AUOptions=4\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic security updates is compliant\", \"finding\": \"The device reports AUOptions=4, which meets the requirement (automatic).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Gemini, Otter\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: Gemini, Otter. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: on\nREQUIRED: off\nDETAIL: A connected computer can read and change data on the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Non-Compliant\", \"title\": \"USB debugging off: does not meet the baseline\", \"finding\": \"Found: on. Required: off. USB debugging lets a connected computer install apps and read data without the usual checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Anyone who can plug the unlocked phone into a computer can copy data or install software.\"}, \"recommendation\": [\"Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; USB debugging should be off.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: >= 6 months for high-risk AI\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"NotApplicable\", \"title\": \"AI usage log retention: not applicable\", \"finding\": \"This check does not apply: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: File History (jdoe)\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Compliant\", \"title\": \"Backup configured: meets the baseline\", \"finding\": \"Found: File History (jdoe). This satisfies the baseline requirement of backup configured.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: anydesk, ssh\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"Remote access services restricted is anydesk, ssh instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: whoopsie error reporting on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Diagnostic data minimised\", \"finding\": \"The device reports whoopsie error reporting on, but the baseline requires usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted is non-compliant\", \"finding\": \"The user reports assistant screen access on, but the baseline requires screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted: does not meet the baseline\", \"finding\": \"Apple Intelligence external integrations restricted is not restricted instead of restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Hugging Face models, Alpaca (Ollama client)\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Unapproved AI tools (shadow AI) is Hugging Face models, Alpaca (Ollama client) instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Password history is non-compliant\", \"finding\": \"Password history is not set instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with pinHistory = 24.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Compliant\", \"title\": \"Record of processing activities: meets the baseline\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 99999\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"Password expiration is PASS_MAX_DAYS 99999 instead of 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Human oversight for high-risk AI is partial instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Removable media encryption is USB storage writable without enforced encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"Firewall is no active firewall instead of active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: backup configured\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Error\", \"title\": \"Backup configured could not be read\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength: does not meet the baseline\", \"finding\": \"Passcode strength is 4-digit code instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 226 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-09 Cumulative Update for Windows 11 (2026-02-17)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Security updates installed recently is 226 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised: does not meet the baseline\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Install unknown apps blocked: does not meet the baseline\", \"finding\": \"Install unknown apps blocked is unknown apps allowed for at least one app instead of all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Perplexity, ChatGPT, Claude\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"The device reports Perplexity, ChatGPT, Claude, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1930 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 653 packages checked against rhel:9 advisories. kernel-debug-devel 0:5.14.0-427.31.1.el9_4 (1884 CVEs, high, actively exploited; fixed in 0:5.14.0-687.54.1.el9_8; e.g. CVE-2019-25162, CVE-2020-10135); kernel-rt-debug-kvm 0:5.14.0-570.42.2.el9_6 (33 CVEs, high, actively exploited; fixed in 0:5.14.0-570.60.1.el9_6; e.g. CVE-2022-50087, CVE-2025-22026); qemu-kvm-common-debuginfo 17:6.2.0-11.el9_0.3 (11 CVEs, high; fixed in 17:10.1.0-17.el9_8; e.g. CVE-2021-3507, CVE-2021-3611); gstreamer1-plugins-base-debuginfo 0:1.22.12-4.el9 (9 CVEs, high; fixed in 0:1.22.12-8.el9_8.2; e.g. CVE-2026-18297, CVE-2026-85150); autocorr-tr 1:7.1.8.1-12.el9_3 (7 CVEs, high; fixed in 1:7.1.8.1-15.el9_8.2; e.g. CVE-2023-6185, CVE-2023-6186); and 9 more. 59 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 1930 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"The device reports USB drives writable without encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 14 (android)\nHOSTNAME: sales-galaxy-44\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=19, Non-Compliant=1, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 2 detected by the browser, 20 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 5 minutes\n- mobile.passcode_strength | Screen lock strength | Compliant | 6-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 14 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-08 (2 months old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Compliant | Android 14\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Meta AI\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Meta AI\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Meta AI\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- aiact.unapproved: Medium\nSCORE (computed): compliant=19, assessed=20, percent=95\nTOP GAPS (computed, highest risk first):\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | ChatGPT, Meta AI\nFIX GUIDANCE (highest-risk gaps):\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it."}, {"role": "assistant", "content": "{\"device\": \"sales-galaxy-44 - Android 14\", \"score\": {\"compliant\": 19, \"assessed\": 20, \"percent\": 95}, \"counts\": {\"Compliant\": 19, \"Non-Compliant\": 1, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"ChatGPT, Meta AI\"}], \"summary\": \"sales-galaxy-44 meets 19 of 20 assessed checks (95%). There are 1 gaps, none of them rated high risk. 10 items are pending. 20 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 179 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 837 packages checked against rhel:8 advisories. glibc-langpack-ber 0:2.28-151.el8 (16 CVEs, critical, actively exploited; fixed in 0:2.28-251.el8_10.22; e.g. CVE-2021-27645, CVE-2021-33574); ipa-server-trust-ad 0:4.8.7-12.module+el8.3.0+8222+c1bff54a (11 CVEs, critical, actively exploited; fixed in 0:4.9.13-20.module+el8.10.0+23534+744f3864; e.g. CVE-2020-11023, CVE-2020-17049); polkit-libs-debuginfo 0:0.115-11.el8_4.1 (1 CVE, high, actively exploited; fixed in 0:0.115-13.el8_5.1; e.g. CVE-2021-4034); podman-docker 3:4.6.1-8.module+el8.9.0+21243+a586538b (73 CVEs, high; fixed in 4:4.9.4-32.module+el8.10.0+24482+e50f4e4f; e.g. CVE-2018-25091, CVE-2021-33198); grub2-tools-efi-debuginfo 1:2.02-78.el8_1.1 (30 CVEs, high; fixed in 1:2.02-170.el8_10.1; e.g. CVE-2020-10713, CVE-2020-14308); and 12 more. 35 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"Known vulnerabilities in installed software is 179 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"Breach response procedure is no instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 56 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last package upgrade: 2026-08-06  17:02:12\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (linux):\n- Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"The device reports 56 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run apt update && apt upgrade (RHEL: dnf upgrade) and check the automatic update logs for errors.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: block\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Error\", \"title\": \"Domain inbound: evidence missing\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: NTP service no, synchronised no\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation is non-compliant\", \"finding\": \"Time synchronisation is NTP service no, synchronised no instead of time synchronised. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"timedatectl set-ntp true (or enable chronyd).\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an enabled administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"Found: none of the approved accounts is an enabled administrator. Required: itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jailbroken\nREQUIRED: not jailbroken\nDETAIL: iOS security protections are bypassed on a jailbroken phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not jailbroken is non-compliant\", \"finding\": \"The user reports jailbroken, but the baseline requires not jailbroken. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 44 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"The device reports 44 days ago, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: jailbroken\nREQUIRED: not jailbroken\nDETAIL: iOS security protections are bypassed on a jailbroken phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Not jailbroken\", \"finding\": \"The user reports jailbroken, but the baseline requires not jailbroken. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LAPTOP-ZRCPWCK\\akaya\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts is non-compliant\", \"finding\": \"Unauthorized employee local administrator accounts is LAPTOP-ZRCPWCK\\\\akaya instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised is non-compliant\", \"finding\": \"Analytics sharing minimised is analytics sharing on instead of analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LAPTOP-RA4F7UO\\asmith\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee local administrator accounts\", \"finding\": \"Found: LAPTOP-RA4F7UO\\\\asmith. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-09 Cumulative Update for Windows 11 Version 22H2 for x64-based Systems (KB5044510).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"The device reports 1 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] How to run an audit of a computer: A web page cannot read a computer's settings, so computers are audited with the free Devseis desktop app for Mac, Windows (beta) and Linux. Download it from the Download page, install it, click Run audit and enter the computer password once. The first run downloads the AI auditor; the audit then takes about 5 to 15 minutes and ends with a PDF report.\n[2] Who is Arx?: Arx is the name of the Devseis security assistant and endpoint auditor (from the Latin for fortress). Arx answers questions about securing computers and phones, ISO 27001, GDPR and the EU AI Act, and audits devices: computers with the Devseis Endpoint Auditor desktop app, phones with the phone check in the browser. Arx runs on the device; nothing is uploaded.\n[3] USB debugging off on Android: USB debugging off on Android. Why it matters: USB debugging lets a connected computer install apps and read data without the usual checks. Risk if missing: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software. Requirement: Developer options / USB debugging are off. How to fix on Android: (1) Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely. How to verify: Re-run the phone check; USB debugging should be off. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] What the Devseis Endpoint Auditor is: The Devseis Endpoint Auditor is a free, open-source, read-only check of a computer or phone against ISO 27001, GDPR technical controls and the EU AI Act. It reads security settings, compares them with a baseline, matches installed software against known vulnerabilities, and an AI auditor that runs on the device writes a report with findings, risk levels and fix steps. It is not a certification and not legal advice.\n\nMY AUDIT: none\n\nQUESTION: Is my computer compliant with ISO 27001?"}, {"role": "assistant", "content": "I can't see your device from here, so I can't tell you yet. Run an audit with the free Devseis desktop app: it reads the settings, changes nothing, and takes about 5 to 15 minutes [1]. Then ask me again and I will answer from your real results."}], "meta": {"task": "chat", "os": "any", "status": "noaudit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection (encryption) [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off (no passcode)\nREQUIRED: on\nDETAIL: Without a passcode, iOS data protection is not active.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (ios):\n- Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Data protection (encryption) is non-compliant\", \"finding\": \"Data protection (encryption) is off (no passcode) instead of on. Full-disk encryption keeps data unreadable if the device is lost or stolen. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Set a passcode (Settings › Face ID & Passcode): iOS data protection encrypts the phone once a passcode is set.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker [encryption.feature] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.feature\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker\", \"finding\": \"Found: off. Required: on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.feature", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup: does not meet the baseline\", \"finding\": \"iCloud Backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 27 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 22 applications inventoried, 14 recognised (31 common applications covered). 7-Zip 24.08 (x64) 24.08 (8 CVEs, high, actively exploited; fixed in 26.02; e.g. CVE-2025-0411, CVE-2025-53816); GIMP 2.8.18 (11 CVEs, critical; fixed in 3.2.0; e.g. CVE-2009-0723, CVE-2009-0733); VLC media player 3.0.9 (6 CVEs, critical; fixed in 3.0.20; e.g. CVE-2020-13428, CVE-2020-26664); Microsoft Teams 2020-10-29 (2 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2025-49737, CVE-2025-53783). 15 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"The device reports 27 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"ClamAV malware protection service is non-compliant\", \"finding\": \"ClamAV malware protection service is not running instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_faillock not enabled\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold is non-compliant\", \"finding\": \"Found: pam_faillock not enabled. Required: 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\", \"Set deny = 5 in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not covered\nREQUIRED: <= 0 critical/high\nDETAIL: Ubuntu 20.04.6 LTS is not in the vulnerability data (covered: ubuntu:22.04, ubuntu:24.04, ubuntu:25.10, ubuntu:26.04, ubuntu:23.10, ubuntu:24.10, ubuntu:25.04, debian:12, debian:13, debian:14, debian:11, rhel:8, rhel:9, rhel:10, almalinux:8, almalinux:9, almalinux:10, rocky:8, rocky:9, rocky:10).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Pending\", \"title\": \"Known vulnerabilities in installed software: decision pending\", \"finding\": \"Found: not covered. Ubuntu 20.04.6 LTS is not in the vulnerability data (covered: ubuntu:22.04, ubuntu:24.04, ubuntu:25.10, ubuntu:26.04, ubuntu:23.10, ubuntu:24.10, ubuntu:25.04, debian:12, debian:13, debian:14, debian:11, rhel:8, rhel:9, rhel:10, almalinux:8, almalinux:9, almalinux:10, rocky:8, rocky:9, rocky:10).\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Auto-Lock on iPhone: Auto-Lock on iPhone. Why it matters: The shorter the idle time before locking, the smaller the window for misuse. Risk if missing: Medium - A long or missing timeout leaves unattended devices open. Requirement: Idle time before lock (including any grace delay) is 1..SCREEN_LOCK_MAX_SECONDS seconds (baseline: 300 = 5 minutes). Phones: Auto-Lock / screen timeout. How to fix on iPhone: (1) Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never). How to verify: Re-run the audit; Screen lock timeout should be 300 seconds or less. Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.7.7 (Clear desk and clear screen); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] AI usage log retention on iPhone: AI usage log retention on iPhone. Why it matters: Deployers of high-risk AI systems must keep the system's logs for at least six months. Risk if missing: Medium - Missing logs prevent monitoring and investigation of high-risk AI use. Requirement: Logs kept >= 6 months if a high-risk AI system is used. Pending until classification; NotApplicable when no AI tool is found. How to fix on iPhone: (1) If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months. How to verify: Log retention is documented for each high-risk AI system. Related controls: ISO 27001:2022 A.8.15 (Logging); EU AI Act Art. 26(6) (Deployers: keep automatically generated logs for at least six months).\n[3] How the phone check works: Phones are checked in the browser: open the Devseis page on the iPhone or Android phone and start the phone check. The browser detects what it can (system version, whether a screen lock is set up) and asks the rest with help on where to look in Settings. Every result is labelled detected or self-reported. The report can be saved or shared as a PDF.\n[4] Msty (AI tool): Msty: Desktop AI client, provider Desktop app (local and cloud models), data location: Local, or the cloud provider of any connected model. Default EU AI Act risk class for general office use: minimal. GDPR: Treat connected cloud models like the corresponding cloud assistant. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n\nMY AUDIT: none\n\nMY DEVICE: iOS 26.0 (ios) on an iPhone, Chrome 141 (iOS); detected by the browser, security settings not checked.\n\nQUESTION: Do I have any viruses?"}, {"role": "assistant", "content": "I can see this is iOS 26.0 on an iPhone, but a web page can't read its security settings, so I can't tell you yet. Run the phone check: open this page on the phone and tap Start the check [3]. Then ask me again and I will answer from your real results."}], "meta": {"task": "chat", "os": "ios", "status": "noaudit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: not configured\nREQUIRED: required diagnostic data only (0-1)\nDETAIL: Windows may send optional diagnostic data.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"Diagnostic data minimised is not configured instead of required diagnostic data only (0-1). Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Ubuntu 20.04.6 LTS\nREQUIRED: Ubuntu 22.04 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"The device reports Ubuntu 20.04.6 LTS, but the baseline requires Ubuntu 22.04 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: chromeremotedesktop, rdp\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is chromeremotedesktop, rdp instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Found: USB storage writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age is non-compliant\", \"finding\": \"Found: 0 days. Required: >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Removable media encryption is USB storage writable without enforced encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"AI register is partial instead of yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (macos):\n- System Settings > Lock Screen: set the screen saver and display-off times and 'Require password' = Immediately, or enforce with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock: meets the baseline\", \"finding\": \"Screen lock is set to on, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 16 (android)\nHOSTNAME: hr-pixel-07\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=14, Non-Compliant=6, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 2 minutes\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | 4-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 16 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-08 (2 months old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Compliant | Android 16\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Non-Compliant | usage & diagnostics on\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Claude\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Claude\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Claude\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.find_my_device: Medium\n- gdpr.auto_updates: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=14, assessed=20, percent=70\nTOP GAPS (computed, highest risk first):\n- gdpr.auto_updates | Automatic system updates | High | manual\n- mobile.passcode_strength | Screen lock strength | Medium | 4-digit PIN\n- mobile.find_my_device | Find My Device | Medium | off\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | ChatGPT, Claude\n- aiact.os_ai_features | AI assistant access restricted | Medium | assistant screen access on\nFIX GUIDANCE (highest-risk gaps):\n- gdpr.auto_updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker).\n- mobile.passcode_strength: Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\n- mobile.find_my_device: Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on."}, {"role": "assistant", "content": "{\"device\": \"hr-pixel-07 - Android 16\", \"score\": {\"compliant\": 14, \"assessed\": 20, \"percent\": 70}, \"counts\": {\"Compliant\": 14, \"Non-Compliant\": 6, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Screen lock strength\", \"risk\": \"Medium\", \"found\": \"4-digit PIN\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My Device\", \"risk\": \"Medium\", \"found\": \"off\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"ChatGPT, Claude\"}, {\"check_id\": \"aiact.os_ai_features\", \"name\": \"AI assistant access restricted\", \"risk\": \"Medium\", \"found\": \"assistant screen access on\"}], \"summary\": \"hr-pixel-07 meets 14 of 20 assessed checks (70%). The most serious gaps are Automatic system updates. 10 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Automatic system updates: Settings › System › Software updates: turn on automatic download (names vary by phone maker).\", \"Screen lock strength: Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\", \"Find My Device: Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: teamviewer, chromeremotedesktop\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"The device reports teamviewer, chromeremotedesktop, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: password-protected screen saver on for all signed-in users\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock matches the required setting\", \"finding\": \"The device reports password-protected screen saver on for all signed-in users, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Never\nREQUIRED: When Unlocked or Never\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Compliant\", \"title\": \"Notification previews hidden when locked is compliant\", \"finding\": \"Found: Never. This satisfies the baseline requirement of When Unlocked or Never. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"The device reports never, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: required\nREQUIRED: password required\nDETAIL: Users checked: sali, jdoe\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement is compliant\", \"finding\": \"The device reports required, which meets the requirement (password required).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: App Store only\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer).\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Pending\", \"title\": \"Apps only from the App Store: needs a follow-up step\", \"finding\": \"The user could not find or confirm this setting. Current state: not answered. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"ClamAV malware protection service is non-compliant\", \"finding\": \"Found: not running. Required: running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"The device reports share Mac analytics=1, share with app developers=0, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: LM Studio, Ollama\nREQUIRED: inventory taken\nDETAIL: Local models: /Users/jdoe/.ollama/models (51G).\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (macos):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"The device reports LM Studio, Ollama, which meets the requirement (inventory taken).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock is non-compliant\", \"finding\": \"Found: Auto-Lock Never. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock: does not meet the baseline\", \"finding\": \"The user reports Auto-Lock Never, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not encrypted\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"Found: not encrypted. Required: recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"Backup configured is none found instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Auto-Lock 5 minutes\nREQUIRED: <= 300 seconds\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Auto-Lock: meets the baseline\", \"finding\": \"Auto-Lock is set to Auto-Lock 5 minutes, in line with the required <= 300 seconds. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (macos):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery does not apply here\", \"finding\": \"Not applicable on this device: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apple Intelligence external integrations restricted\", \"finding\": \"The user reports ChatGPT extension on, but the baseline requires external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: NTP service yes, synchronised yes\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation is compliant\", \"finding\": \"Found: NTP service yes, synchronised yes. This satisfies the baseline requirement of time synchronised.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; screen saver off\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"The device reports no machine inactivity limit; screen saver off, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 10 (android)\nHOSTNAME: dev-galaxy-37\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=15, Non-Compliant=4, NotApplicable=0, Pending=11, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 15 seconds\n- mobile.passcode_strength | Screen lock strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Pending | not answered\n- encryption.enabled | Device encryption | Compliant | on (Android 10 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-09 (1 month old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 10\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | sensitive content hidden\n- aiact.inventory | AI software inventory | Compliant | Microsoft Copilot\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Microsoft Copilot\n- aiact.risk_classification | AI system risk classification | Pending | Microsoft Copilot\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- aiact.unapproved: Medium\nSCORE (computed): compliant=15, assessed=19, percent=79\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- gdpr.os_supported | Supported operating system version | High | Android 10\n- gdpr.backup_configured | Google backup | High | no backup\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | Microsoft Copilot\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- gdpr.os_supported: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\n- gdpr.backup_configured: Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps."}, {"role": "assistant", "content": "{\"device\": \"dev-galaxy-37 - Android 10\", \"score\": {\"compliant\": 15, \"assessed\": 19, \"percent\": 79}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 4, \"NotApplicable\": 0, \"Pending\": 11, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 10\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Google backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"Microsoft Copilot\"}], \"summary\": \"dev-galaxy-37 meets 15 of 19 assessed checks (79%). The most serious gaps are Google Play Protect, Supported operating system version and Google backup. 11 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Supported operating system version: Update to Android 13 or later in Settings › System › Software updates; if the maker no longer offers it, replace the phone.\", \"Google backup: Settings › Google › Backup: turn on Backup by Google One (or the employer's approved backup); keep work data only in company-approved apps.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rooted or bootloader unlocked\nREQUIRED: not rooted\nDETAIL: Android security protections are weakened on a rooted phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not rooted: does not meet the baseline\", \"finding\": \"Not rooted is rooted or bootloader unlocked instead of not rooted. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in ArqAgent.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: decision pending\", \"finding\": \"Found: date not readable. Confirm the last backup date in ArqAgent.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: XProtect and Gatekeeper on\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (macos):\n- Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\nverify: Re-run the audit; the malware protection service should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Error\", \"title\": \"Apple malware protection service could not be read\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker protection [encryption.protection] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: protection Off, None\nREQUIRED: protection on, AES\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Encryption only protects the data when it is active, fully applied and uses a strong cipher.\nrisk: High - Suspended or partial encryption leaves data readable on a lost device.\nfix (windows):\n- Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\nverify: Re-run the audit; Disk encryption protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.protection\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker protection: does not meet the baseline\", \"finding\": \"BitLocker protection is protection Off, None instead of protection on, AES. Encryption only protects the data when it is active, fully applied and uses a strong cipher.\", \"risk\": {\"level\": \"High\", \"reason\": \"Suspended or partial encryption leaves data readable on a lost device.\"}, \"recommendation\": [\"Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Disk encryption protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.protection", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Android security update [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: security patch 2025-03 (19 months old)\nREQUIRED: <= 90 days old\nDETAIL: Security fixes published since 2025-03 are missing; Google publishes Android fixes every month.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (android):\n- Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Android security update: does not meet the baseline\", \"finding\": \"The user reports security patch 2025-03 (19 months old), but the baseline requires <= 90 days old. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: EncryptionInProgress (42%)\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption is non-compliant\", \"finding\": \"Encryption is EncryptionInProgress (42%) instead of fully encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"The device reports no lockout configured, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutwindow:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: ssh\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"The device reports ssh, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Compliant\", \"title\": \"Record of processing activities is compliant\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 44 days old\nREQUIRED: <= 3 days old\nDETAIL: /var/lib/clamav/daily.cld\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures is non-compliant\", \"finding\": \"Found: 44 days old. Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: >= 6 months for high-risk AI\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"NotApplicable\", \"title\": \"AI usage log retention does not apply here\", \"finding\": \"Not applicable on this device: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-10 Cumulative Update for Windows 11 Version 24H2 for x64-based Systems (KB5041765), Security Intelligence Update for Microsoft Defender Antivirus - KB2267602 (Version 1.411.0).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Found: 2 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: required\nREQUIRED: password required\nDETAIL: Users checked: mkhan\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement is compliant\", \"finding\": \"Screen lock password requirement is set to required, in line with the required password required.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked: does not meet the baseline\", \"finding\": \"The user reports Always, but the baseline requires When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password history\", \"finding\": \"Password history is 0 instead of >= 24. Without a password history, users can return to old, possibly exposed passwords.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Reused passwords keep compromised credentials valid.\"}, \"recommendation\": [\"Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\", \"Or set remember = 24 in /etc/security/pwhistory.conf where supported.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password history should read 24 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Pending\", \"title\": \"Passcode strength is pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"The device reports 1 key slots, 0 unlock tokens, but the baseline requires recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: none; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging: does not meet the baseline\", \"finding\": \"The device reports logon: none; account management: none, but the baseline requires logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates: does not meet the baseline\", \"finding\": \"The device reports freshclam not running, but the baseline requires automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age is non-compliant\", \"finding\": \"The device reports 0 days, but the baseline requires >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Public inbound: does not meet the baseline\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock is non-compliant\", \"finding\": \"Auto-Lock is Auto-Lock Never instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: alphanumeric\nREQUIRED: >= 6 digits or alphanumeric\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Compliant\", \"title\": \"Passcode strength matches the required setting\", \"finding\": \"Passcode strength is set to alphanumeric, in line with the required >= 6 digits or alphanumeric. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: only: none\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Error\", \"title\": \"Unapproved AI tools (shadow AI) could not be read\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically is non-compliant\", \"finding\": \"Found: set by hand. Required: set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 259 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 16 applications inventoried, 11 recognised (23 common applications covered). Adobe Acrobat Reader 25.001.20693 (42 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-64785, CVE-2025-64899); Firefox 152.0.6 (201 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100768, CVE-2026-100779); zoom.us 6.4.0 (9 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-46788, CVE-2025-49460); iTerm 3.4.18 (5 CVEs, critical; fixed in 3.4.20; e.g. CVE-2023-46300, CVE-2023-46301); LibreOffice 24.2.4 (2 CVEs, high; fixed in 24.2.5.1; e.g. CVE-2024-6472, CVE-2024-7788). 72 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 259 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Encryption on Windows: Encryption on Windows. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: The system disk is fully encrypted (BitLocker / LUKS / FileVault). Phones: iOS data protection is on when a passcode is set; Android 10+ encrypts by default. How to fix on Windows: (1) Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[2] BitLocker on Windows: BitLocker on Windows. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: BitLocker is turned on for the system drive. How to fix on Windows: (1) Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[3] Removable media encryption on Windows: Removable media encryption on Windows. Why it matters: Personal data copied to removable media must be encrypted, or copying must be blocked. Risk if missing: Medium - A lost unencrypted USB drive with personal data is a reportable breach. Requirement: Data cannot be written to unencrypted removable drives. How to fix on Windows: (1) Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access. How to verify: Re-run the audit; Removable media encryption should be Compliant. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption); GDPR Art. 34 (Communication of a personal data breach to the data subject).\n[4] BitLocker protection on Windows: BitLocker protection on Windows. Why it matters: Encryption only protects the data when it is active, fully applied and uses a strong cipher. Risk if missing: High - Suspended or partial encryption leaves data readable on a lost device. Requirement: Encryption protection is active (not suspended or in progress) with a strong cipher. How to fix on Windows: (1) Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method. How to verify: Re-run the audit; Disk encryption protection should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n\nMY AUDIT: none\n\nQUESTION: How do I fix bitlocker protection on Windows?"}, {"role": "assistant", "content": "Encryption only protects the data when it is active, fully applied and uses a strong cipher. To fix it on Windows [4]:\n1. Resume protection if it is suspended (Resume-BitLocker -MountPoint C:) and let encryption finish; use an AES-XTS method.\n\nTo check it worked: Re-run the audit; Disk encryption protection should be Compliant."}], "meta": {"task": "chat", "os": "windows", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Windows Update service disabled\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates is non-compliant\", \"finding\": \"The device reports Windows Update service disabled, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows Update > Configure Automatic Updates = 4 (auto download and schedule the install), or use Windows Update for Business / Intune; make sure the Windows Update service is not disabled.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: When Unlocked\nREQUIRED: When Unlocked or Never\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Compliant\", \"title\": \"Notification previews hidden when locked is compliant\", \"finding\": \"Notification previews hidden when locked is set to When Unlocked, in line with the required When Unlocked or Never. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Found: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after off seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"Found: password after off seconds. Required: password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training matches the required setting\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 20\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (windows):\n- Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n- For a stand-alone device: net accounts /lockoutthreshold:5\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold: does not meet the baseline\", \"finding\": \"Account lockout threshold is 20 instead of 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\", \"For a stand-alone device: net accounts /lockoutthreshold:5\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 2 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: 2026-09 Cumulative Update for Windows 11 Version 24H2 for x64-based Systems (KB5041197), Security Intelligence Update for Microsoft Defender Antivirus - KB2267602 (Version 1.486.0).\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (windows):\n- Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"The device reports 2 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install the pending updates now (Settings > Windows Update > Install all), restart if asked, and check that automatic updates are not paused.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 90 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration is compliant\", \"finding\": \"Found: 90 days. This satisfies the baseline requirement of 1-365 days.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 99999\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"Found: PASS_MAX_DAYS 99999. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: openssl-libs, sudo, kernel. Uses the cached repository data.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Found: 15 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 64 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Recent backup\", \"finding\": \"The device reports 64 days ago, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000161, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 27.0 (ios)\nHOSTNAME: SALES-iPhone-24\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=17, Non-Compliant=0, NotApplicable=2, Pending=10, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 30 seconds\n- mobile.passcode_strength | Passcode strength | Compliant | 6-digit code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store and employer-managed apps\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Pending | not answered\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 27.0\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Pending | not answered\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\nSCORE (computed): compliant=17, assessed=17, percent=100\nTOP GAPS (computed, highest risk first):\nFIX GUIDANCE (highest-risk gaps):"}, {"role": "assistant", "content": "{\"device\": \"SALES-iPhone-24 - iOS 27.0\", \"score\": {\"compliant\": 17, \"assessed\": 17, \"percent\": 100}, \"counts\": {\"Compliant\": 17, \"Non-Compliant\": 0, \"NotApplicable\": 2, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [], \"summary\": \"SALES-iPhone-24 meets 17 of 17 assessed checks (100%). No gaps were found. 10 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Found: not enforced. Required: enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 26.5.2 (macos)\nAI TOOLS FOUND: Claude, Cursor, Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\n- Cursor: AI code editor; provider Anysphere (US); data: Code sent to the provider and its model providers unless privacy mode is enforced; default risk minimal; GDPR: Enforce privacy mode through a business plan and sign a data processing agreement.\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Cursor\", \"kind\": \"AI code editor\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Enforce privacy mode through a business plan and sign a data processing agreement.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude, Cursor, Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: automatic app updates on\nREQUIRED: automatic app updates on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Compliant\", \"title\": \"App updates: meets the baseline\", \"finding\": \"Found: automatic app updates on. This satisfies the baseline requirement of automatic app updates on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Real-time protection [malware.realtime] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Real-time protection scans files and processes as they are used.\nrisk: High - Without it, malware is only found by scheduled scans, after it has run.\nfix (windows):\n- Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\nverify: Re-run the audit; Real-time protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.realtime\", \"status\": \"Non-Compliant\", \"title\": \"Real-time protection is non-compliant\", \"finding\": \"Found: off. Required: on. Real-time protection scans files and processes as they are used.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without it, malware is only found by scheduled scans, after it has run.\"}, \"recommendation\": [\"Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Real-time protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.realtime", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 61 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 21 applications inventoried, 10 recognised (23 common applications covered). Adobe Acrobat Reader 25.001.20432 (49 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-43550, CVE-2025-43573); zoom.us 6.5.0 (6 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-62483, CVE-2025-64739); VLC 3.0.19 (2 CVEs, critical; fixed in 3.0.20; e.g. CVE-2023-47359, CVE-2023-47360); Safari 26.6 (3 CVEs, high; fixed in 26.6.1; e.g. CVE-2026-43794, CVE-2026-65390); iTerm 3.5.2 (1 CVE, high; e.g. CVE-2026-41253). 50 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"The device reports 61 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Real-time protection [malware.realtime] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Real-time protection scans files and processes as they are used.\nrisk: High - Without it, malware is only found by scheduled scans, after it has run.\nfix (windows):\n- Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\nverify: Re-run the audit; Real-time protection should be on."}, {"role": "assistant", "content": "{\"check_id\": \"malware.realtime\", \"status\": \"Non-Compliant\", \"title\": \"Real-time protection is non-compliant\", \"finding\": \"Found: off. Required: on. Real-time protection scans files and processes as they are used.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without it, malware is only found by scheduled scans, after it has run.\"}, \"recommendation\": [\"Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Real-time protection should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.realtime", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: guest login 1, guest file sharing 0\nREQUIRED: guest login off\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (macos):\n- sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Guest account disabled is non-compliant\", \"finding\": \"Found: guest login 1, guest file sharing 0. Required: guest login off. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: too long for: pmartin (1800s)\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: pmartin, mkhan\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Screen lock timeout is too long for: pmartin (1800s) instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout threshold\", \"finding\": \"The device reports not set, but the baseline requires 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxFailedAttempts = 5.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"Found: none found. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: EncryptionInProgress (42%)\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Encryption\", \"finding\": \"Encryption is EncryptionInProgress (42%) instead of fully encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Standard (pmartin)\nREQUIRED: Standard\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Compliant\", \"title\": \"Signed-in user account type matches the required setting\", \"finding\": \"Found: Standard (pmartin). This satisfies the baseline requirement of Standard.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device: does not meet the baseline\", \"finding\": \"The user reports off, but the baseline requires on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"Human oversight for high-risk AI is partial instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000661, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: download and install on\nDETAIL: Security updates are not installed automatically.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic iOS updates\", \"finding\": \"Found: off. Required: download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: EncryptionInProgress (42%)\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption is non-compliant\", \"finding\": \"Found: EncryptionInProgress (42%). Required: fully encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Compliant\", \"title\": \"Data processing agreements is compliant\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: download and install on\nREQUIRED: download and install on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic iOS updates is compliant\", \"finding\": \"Automatic iOS updates is set to download and install on, in line with the required download and install on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nHOSTNAME: ops-ubuntu-60\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=20, Non-Compliant=30, NotApplicable=2, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Non-Compliant | 12\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | Non-Compliant | PASS_MIN_DAYS 0\n- password.complexity | Password complexity | Compliant | minclass=3, required classes=0\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Non-Compliant | not set\n- lockout.observation | Lockout observation period | Non-Compliant | not set\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Non-Compliant | no signature database\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Non-Compliant | no active firewall\n- firewall.inbound | Firewall inbound policy | Non-Compliant | allow (no firewall)\n- firewall.outbound | Firewall outbound policy | Non-Compliant | no firewall\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | mkhan\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (rschmit)\n- encryption.enabled | Encryption | Non-Compliant | root filesystem not encrypted\n- encryption.protection | Disk encryption protection | Non-Compliant | not encrypted\n- encryption.recovery | Disk encryption recovery | Non-Compliant | not encrypted\n- vuln.pending_updates | Pending security updates | Non-Compliant | 18 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 5 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 9 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 22.04.5 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | smb, rdp\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | popularity-contest on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | LM Studio\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | LM Studio\n- aiact.risk_classification | AI system risk classification | Pending | LM Studio\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.min_age: Low\n- lockout.duration: Low\n- lockout.observation: Low\n- malware.antivirus: High\n- malware.signature_updates: Medium\n- malware.signatures: High\n- malware.on_access: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- firewall.outbound: Low\n- usb.write: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.ropa: High\n- org.breach_procedure: High\n- org.ai_register: Medium\n- org.ai_literacy: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=20, assessed=50, percent=40\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.signatures | Malware signatures | High | no signature database\n- firewall.enabled | Firewall | High | no active firewall\n- firewall.inbound | Firewall inbound policy | High | allow (no firewall)\n- admin.unauthorized | Unauthorized employee administrator accounts | High | mkhan\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\n- firewall.enabled: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld."}, {"role": "assistant", "content": "{\"device\": \"ops-ubuntu-60 - Ubuntu 22.04.5 LTS 22.04\", \"score\": {\"compliant\": 20, \"assessed\": 50, \"percent\": 40}, \"counts\": {\"Compliant\": 20, \"Non-Compliant\": 30, \"NotApplicable\": 2, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.signatures\", \"name\": \"Malware signatures\", \"risk\": \"High\", \"found\": \"no signature database\"}, {\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"no active firewall\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"allow (no firewall)\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"mkhan\"}], \"summary\": \"ops-ubuntu-60 meets 20 of 50 assessed checks (40%). The most serious gaps are Antivirus, Malware signatures and Firewall. 4 items are pending.\", \"next_steps\": [\"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"Malware signatures: systemctl enable --now clamav-freshclam, or run freshclam once to update now.\", \"Firewall: Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 0 pending\nREQUIRED: <= 0 pending\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (macos):\n- Install the pending updates: System Settings > General > Software Update, or softwareupdate -ia, then restart if asked.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Compliant\", \"title\": \"Pending security updates: meets the baseline\", \"finding\": \"Found: 0 pending. This satisfies the baseline requirement of <= 0 pending.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Current state: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates is non-compliant\", \"finding\": \"Automatic security updates is check=1 download=1 security=1 macOS-install=0 instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (windows):\n- Enable real-time protection in Microsoft Defender (on-access protection follows it).\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Compliant\", \"title\": \"On-access protection is compliant\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training: does not meet the baseline\", \"finding\": \"AI literacy training is no instead of yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: guest login 1, guest file sharing 0\nREQUIRED: guest login off\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (macos):\n- sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Guest account disabled is non-compliant\", \"finding\": \"Guest account disabled is guest login 1, guest file sharing 0 instead of guest login off. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 241 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 14 applications inventoried, 11 recognised (23 common applications covered). Safari 18.6 (28 CVEs, critical, actively exploited; fixed in 26.6.1; e.g. CVE-2025-14174, CVE-2025-43342); Adobe Acrobat Reader 25.001.20693 (42 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2025-64785, CVE-2025-64899); Google Chrome 153.0.8010.36 (171 CVEs, critical; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301). 280 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"The device reports 241 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nDETAIL: pam_pwquality not enabled; pam_unix minimum\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password length\", \"finding\": \"Minimum password length is 6 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Cursor, Ollama, Claude\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (macos):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps and restrict installations through your MDM.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"The device reports Cursor, Ollama, Claude, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps and restrict installations through your MDM.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: BitLocker recovery\", \"finding\": \"Found: key protectors: Tpm. Required: recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"Found: partial. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Breach response procedure is partial instead of yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Microsoft Defender Antimalware [malware.service] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: service True, antivirus True\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (windows):\n- Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Compliant\", \"title\": \"Microsoft Defender Antimalware matches the required setting\", \"finding\": \"Microsoft Defender Antimalware is set to service True, antivirus True, in line with the required running.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamonacc running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Compliant\", \"title\": \"On-access malware protection is compliant\", \"finding\": \"Found: clamonacc running. This satisfies the baseline requirement of on-access scanning.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Date & time set automatically\", \"finding\": \"The user reports set by hand, but the baseline requires set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › General › Date & Time: turn on Set Automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period: does not meet the baseline\", \"finding\": \"The device reports no lockout, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Date & time set automatically: does not meet the baseline\", \"finding\": \"The user reports set by hand, but the baseline requires set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"Found: disabled. Required: enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: <= 300 seconds\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Error\", \"title\": \"Screen lock timeout could not be read\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: LM Studio, Ollama\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: LM Studio, Ollama. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 1 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Compliant\", \"title\": \"Minimum password age is compliant\", \"finding\": \"The device reports 1 days, which meets the requirement (>= 1 days).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Security event logging on Linux: Security event logging on Linux. Why it matters: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline. Risk if missing: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork. Requirement: Logon and account-change events are recorded (Windows audit policy, Linux auditd, macOS unified log). How to fix on Linux: (1) Install and start auditd (apt install auditd; systemctl enable --now auditd) and keep the journal persistent (Storage=persistent in /etc/systemd/journald.conf). How to verify: Re-run the audit; Security event logging should be Compliant. Related controls: ISO 27001:2022 A.8.15 (Logging); GDPR Art. 32(1)(d) (Security of processing: regular testing and evaluation); GDPR Art. 33 (Notification of a personal data breach to the supervisory authority).\n[2] How to install on Windows: Run the downloaded setup .exe. Windows SmartScreen shows 'Windows protected your PC' because the app is not yet code-signed: click More info, then Run anyway. Click Next, Install and Finish. When the audit starts, click Yes in the User Account Control prompt. The Windows version is a beta.\n[3] Antivirus on Linux: Antivirus on Linux. Why it matters: Anti-malware software detects and blocks known malicious files and behaviour. Risk if missing: High - Without working anti-malware, ransomware and info-stealers can run unchallenged. Requirement: An enabled, up-to-date antivirus product is present. Android: Google Play Protect scanning is on. How to fix on Linux: (1) Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon. How to verify: Re-run the audit; Antivirus should be Compliant. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Signed-in user account type on Linux: Signed-in user account type on Linux. Why it matters: Day-to-day work under an administrator account lets any malware or mistake change the whole system. Risk if missing: High - Malware started by the signed-in user runs with full administrator rights. Requirement: The everyday signed-in user is a Standard account (not an administrator), unless listed in APPROVED_ADMINS. How to fix on Linux: (1) Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account. How to verify: Re-run the audit; Signed-in user account type should read Standard. Related controls: ISO 27001:2022 A.8.2 (Privileged access rights); ISO 27001:2022 A.5.15 (Access control); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Debian GNU/Linux 12 (bookworm) 12 (linux), audited 2026-10-01; 42 of 49 assessed checks compliant; 7 gaps, 2 pending, 1 not read. Gaps: Antivirus: none running; Malware signature updates: freshclam not running; Malware signatures: 34 days old; Screen lock password requirement: not required for: sali; Outdated third-party applications: 10 outdated; Security event logging: auditd not running, journal volatile; Diagnostic data minimised: whoopsie error reporting on.\n\nQUESTION: Where do I start?"}, {"role": "assistant", "content": "Your audit found 7 gaps. Start with Antivirus (found: none running). Anti-malware software detects and blocks known malicious files and behaviour. To fix it [3]:\n1. Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n\nThen look at: Malware signature updates, Malware signatures, Screen lock password requirement."}], "meta": {"task": "chat", "os": "linux", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"The device reports not set, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 106 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iPadOS 26.5 checked against NVD iOS records. iPadOS 26.5 (106 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-28928). 180 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS is non-compliant\", \"finding\": \"Found: 106 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (DEVSEIS\\asmith)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"Signed-in user account type is Administrator (DEVSEIS\\\\asmith) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Disk encryption protection on Linux: Disk encryption protection on Linux. Why it matters: Encryption only protects the data when it is active, fully applied and uses a strong cipher. Risk if missing: High - Suspended or partial encryption leaves data readable on a lost device. Requirement: Encryption protection is active (not suspended or in progress) with a strong cipher. How to fix on Linux: (1) Use LUKS2 with aes-xts-plain64 and a 512-bit key (AES-256) for new installations; plan migration of weaker setups. How to verify: Re-run the audit; Disk encryption protection should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[2] Encryption on Linux: Encryption on Linux. Why it matters: Full-disk encryption keeps data unreadable if the device is lost or stolen. Risk if missing: High - A lost unencrypted laptop is a personal data breach that may have to be reported. Requirement: The system disk is fully encrypted (BitLocker / LUKS / FileVault). Phones: iOS data protection is on when a passcode is set; Android 10+ encrypts by default. How to fix on Linux: (1) Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup. How to verify: Re-run the audit; Encryption should be Compliant. Related controls: ISO 27001:2022 A.8.24 (Use of cryptography); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption).\n[3] GDPR Art. 28 Processor: GDPR Art. 28 (Processor): Use only processors that give sufficient guarantees, under a written contract with required terms. An auditor looks for: A data processing agreement with each processor. On endpoints: Cloud AI tools used with work data are processors and need an agreement.\n[4] ChatGPT (AI tool): ChatGPT: Cloud AI assistant, provider OpenAI (US), data location: Provider cloud outside the EEA unless an EU data residency plan is used. Default EU AI Act risk class for general office use: minimal. GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n\nMY AUDIT: none\n\nMY DEVICE: Linux (linux) on a Linux computer, Chrome 152; detected by the browser, security settings not checked.\n\nQUESTION: Give me a workout plan."}, {"role": "assistant", "content": "I can only help with device security, ISO 27001, GDPR and the EU AI Act, so I can't answer that. Ask me, for example, how to turn on disk encryption, whether an AI tool is OK for work, or run an audit of this device."}], "meta": {"task": "chat", "os": "any", "status": "oos", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: macOS 13.7.8\nREQUIRED: macOS 14 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (macos):\n- Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"Found: macOS 13.7.8. Required: macOS 14 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 542 source packages checked against ubuntu:22.04 advisories. dotnet7 7.0.107-0ubuntu1~22.04.1 (2 CVEs, high, actively exploited; fixed in 7.0.112-0ubuntu1~22.04.1; e.g. CVE-2023-38180, CVE-2023-44487); git 1:2.34.1-1ubuntu1.12 (1 CVE, high, actively exploited; fixed in 1:2.34.1-1ubuntu1.13; e.g. CVE-2025-48384); openssl 3.0.2-0ubuntu1.9 (2 CVEs, high; fixed in 3.0.2-0ubuntu1.30; e.g. CVE-2026-84782, CVE-2026-45447). 92 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"Known vulnerabilities in installed software is 5 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: machine inactivity limit set\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock matches the required setting\", \"finding\": \"Screen lock is set to machine inactivity limit set, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not enforced\nREQUIRED: enforced\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (macos):\n- Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Found: not enforced. Required: enforced. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with requireAlphanumeric = true (and minComplexChars = 1 if special characters are required).\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 59 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-08-05 19:15:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently is non-compliant\", \"finding\": \"Security updates installed recently is 59 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: set automatically\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (android):\n- Settings › System › Date & time: turn on Set time automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Pending\", \"title\": \"Date & time set automatically: decision pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › System › Date & time: turn on Set time automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Microsoft Defender Antimalware [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: service True, antivirus False\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (windows):\n- Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Microsoft Defender Antimalware is non-compliant\", \"finding\": \"The device reports service True, antivirus False, but the baseline requires running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: key protectors: Tpm\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"BitLocker recovery is non-compliant\", \"finding\": \"Found: key protectors: Tpm. Required: recovery password present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Current state: Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: anydesk\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is anydesk instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"The device reports 5 minutes, but the baseline requires >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutwindow:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 14 (android)\nHOSTNAME: dev-pixel-42\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=23, Non-Compliant=3, NotApplicable=0, Pending=4, Error=0\nEVIDENCE: 2 detected by the browser, 20 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 2 minutes\n- mobile.passcode_strength | Screen lock strength | Compliant | 6-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Compliant | on\n- encryption.enabled | Device encryption | Compliant | on (Android 14 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-08 (2 months old)\n- vuln.outdated_apps | App updates | Pending | not answered\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Compliant | Android 14\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | Google backup | Compliant | phone maker cloud backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Compliant | usage & diagnostics off\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | no notifications on lock screen\n- aiact.inventory | AI software inventory | Compliant | Character.AI, Meta AI, Otter\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Character.AI, Meta AI, Otter\n- aiact.risk_classification | AI system risk classification | Pending | Character.AI, Meta AI, Otter\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=23, assessed=26, percent=88\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | Character.AI, Meta AI, Otter\n- aiact.os_ai_features | AI assistant access restricted | Medium | assistant screen access on\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- aiact.os_ai_features: In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it."}, {"role": "assistant", "content": "{\"device\": \"dev-pixel-42 - Android 14\", \"score\": {\"compliant\": 23, \"assessed\": 26, \"percent\": 88}, \"counts\": {\"Compliant\": 23, \"Non-Compliant\": 3, \"NotApplicable\": 0, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"Character.AI, Meta AI, Otter\"}, {\"check_id\": \"aiact.os_ai_features\", \"name\": \"AI assistant access restricted\", \"risk\": \"Medium\", \"found\": \"assistant screen access on\"}], \"summary\": \"dev-pixel-42 meets 23 of 26 assessed checks (88%). The most serious gaps are Google Play Protect. 4 items are pending. 20 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"AI assistant access restricted: In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Found: no. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"The device reports partial, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: unified logging active\nREQUIRED: security events logged\nDETAIL: macOS unified log records authentication events; retention is managed by the OS.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (macos):\n- Unified logging is built in; forward security logs to your log platform if long retention is required.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging matches the required setting\", \"finding\": \"Security event logging is set to unified logging active, in line with the required security events logged.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Where to find Screen lock strength on Android: Screen lock strength on Android: What kind of screen lock do you use? Where to look: Settings › Security & privacy › Device unlock › Screen lock\n[2] Screen lock strength on Android: Screen lock strength on Android. Why it matters: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Risk if missing: Medium - A weak passcode lets someone who has the phone unlock it and read work data. Requirement: Passcode / PIN has at least MOBILE_PASSCODE_MIN_DIGITS digits, or is alphanumeric. Android pattern or swipe locks do not comply. How to fix on Android: (1) Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern. How to verify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric. Related controls: ISO 27001:2022 A.5.17 (Authentication information); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Sensitive notifications hidden when locked on Android: Sensitive notifications hidden when locked on Android. Why it matters: Previews show message and email content, often personal data, to anyone who sees the locked phone. Risk if missing: Low - Personal data in notifications can be read without unlocking the phone. Requirement: Message and email previews are not shown on the locked screen. How to fix on Android: (1) Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications. How to verify: Re-run the phone check; previews should be hidden when the phone is locked. Related controls: ISO 27001:2022 A.7.7 (Clear desk and clear screen); ISO 27001:2022 A.8.1 (User endpoint devices); GDPR Art. 5(1)(f) (Integrity and confidentiality); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Where to find Screen lock on Android: Screen lock on Android: Which screen lock is set? Where to look: Settings › Security & privacy › Device unlock › Screen lock (names vary by phone maker)\n\nMY AUDIT: none\n\nQUESTION: How do I see screen lock strength on Android?"}, {"role": "assistant", "content": "Go to Settings › Security & privacy › Device unlock › Screen lock [1]"}], "meta": {"task": "chat", "os": "android", "status": "phone", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAI TOOLS FOUND: GPT4All\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- GPT4All: Local model runner; provider Nomic AI (runs locally); data: On the device unless optional data sharing is enabled; default risk minimal; GDPR: Keep the optional data lake sharing switched off.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"GPT4All\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Keep the optional data lake sharing switched off.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: GPT4All.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nHOSTNAME: dev-ubuntu-30\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=35, Non-Compliant=14, NotApplicable=2, Pending=5, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Compliant | 2 days\n- password.complexity | Password complexity | Non-Compliant | pam_pwquality not enabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Non-Compliant | freshclam not running\n- malware.signatures | Malware signatures | Compliant | 1 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | nftables active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | pmartin\n- user.account_type | Signed-in user account type | Compliant | Standard (pmartin)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 95 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Compliant | 26 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Red Hat Enterprise Linux 9.6 (Plow)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | GPT4All\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | GPT4All\n- aiact.risk_classification | AI system risk classification | Pending | GPT4All\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.complexity: Medium\n- malware.antivirus: High\n- malware.signature_updates: Medium\n- malware.on_access: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- vuln.known_vulnerabilities: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.breach_procedure: High\n- org.processor_agreements: High\nSCORE (computed): compliant=35, assessed=49, percent=71\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- admin.unauthorized | Unauthorized employee administrator accounts | High | pmartin\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 95 critical/high\n- gdpr.backup_configured | Backup configured | High | none\n- org.breach_procedure | Breach response procedure | High | partial\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"dev-ubuntu-30 - Red Hat Enterprise Linux 9.6 (Plow) 9.6\", \"score\": {\"compliant\": 35, \"assessed\": 49, \"percent\": 71}, \"counts\": {\"Compliant\": 35, \"Non-Compliant\": 14, \"NotApplicable\": 2, \"Pending\": 5, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"pmartin\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"95 critical/high\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"org.breach_procedure\", \"name\": \"Breach response procedure\", \"risk\": \"High\", \"found\": \"partial\"}], \"summary\": \"dev-ubuntu-30 meets 35 of 49 assessed checks (71%). The most serious gaps are Antivirus, Unauthorized employee administrator accounts and Known vulnerabilities in installed software. 5 items are pending.\", \"next_steps\": [\"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 3 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"The device reports none found, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: EncryptionInProgress (42%)\nREQUIRED: fully encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (windows):\n- Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"Found: EncryptionInProgress (42%). Required: fully encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on BitLocker for the system drive (Enable-BitLocker -MountPoint C: -EncryptionMethod XtsAes256 -TpmProtector), preferably through Intune or Group Policy with key backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: machine inactivity limit set\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (windows):\n- Set Group Policy: Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds, or enforce a password-protected screen saver by Group Policy/Intune.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock matches the required setting\", \"finding\": \"Screen lock is set to machine inactivity limit set, in line with the required on.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 600 idle, password delay 60\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Screen lock timeout is 600 idle, password delay 60 instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength is non-compliant\", \"finding\": \"Passcode strength is 4-digit code instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Unapproved AI tools (shadow AI) on Windows: Unapproved AI tools (shadow AI) on Windows. Why it matters: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Risk if missing: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations. Requirement: Every AI tool found is listed in APPROVED_AI_TOOLS. How to fix on Windows: (1) Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it. (2) Publish an AI acceptable-use policy so staff know which tools are allowed. (3) Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control. How to verify: Re-run the audit; Unapproved AI tools should read none. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 28 (Processor); GDPR Art. 44 (General principle for transfers); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[2] AI software inventory on Windows: AI software inventory on Windows. Why it matters: Deployers must know which AI systems they use before they can classify and govern them. Risk if missing: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools. Requirement: AI apps, local AI services and local model folders are listed (Compliant = inventory taken). How to fix on Windows: (1) Add the AI tools found on this device to the organisation's AI register, with owner and purpose. How to verify: The AI register lists every tool in the inventory. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); GDPR Art. 30 (Records of processing activities); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[3] AI system risk classification on Windows: AI system risk classification on Windows. Why it matters: Each AI system must be checked against the banned practices and the high-risk uses in Annex III. Risk if missing: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards. Requirement: Each AI tool found is classified (prohibited / high-risk / limited / minimal). Done by the auditor model: Pending in the collector output, NotApplicable when no AI tool is found. How to fix on Windows: (1) Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose. How to verify: Every AI tool in the inventory has a classification in the AI register. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); EU AI Act Art. 5 (Prohibited AI practices); EU AI Act Art. 6 (Classification rules for high-risk AI systems); EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)).\n[4] Meta AI (AI tool): Meta AI: Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps), provider Meta (US), data location: Provider cloud outside the EEA. Default EU AI Act risk class for general office use: minimal. GDPR: Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n\nMY AUDIT: none\n\nMY DEVICE: Windows 11 (windows) on a Windows PC (x86-64), Firefox 143; detected by the browser, security settings not checked.\n\nQUESTION: Is Meta AI safe for my company?"}, {"role": "assistant", "content": "Meta AI is a cloud ai assistant (built into whatsapp, instagram and facebook apps) from Meta (US); data goes to: provider cloud outside the eea [4]. For general office use its default EU AI Act risk class is minimal. GDPR: Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\n\nWhether it is allowed depends on your organisation's approved AI tools list and on what you use it for: some purposes (for example screening job applicants) make any AI tool high-risk."}], "meta": {"task": "chat", "os": "any", "status": "tool", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Account lockout duration on Linux: Account lockout duration on Linux. Why it matters: The lockout must last long enough to slow down repeated guessing. Risk if missing: Low - A short lockout lets attackers resume guessing almost at once. Requirement: Locked for >= LOCKOUT_DURATION_MIN_MINUTES minutes, or until an administrator unlocks. How to fix on Linux: (1) Set unlock_time = 900 (seconds) in /etc/security/faillock.conf. How to verify: Re-run the audit; Account lockout duration should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Account lockout threshold on Linux: Account lockout threshold on Linux. Why it matters: Locking an account after a few wrong passwords stops online guessing. Risk if missing: Medium - Unlimited attempts allow brute-force attacks against local accounts. Requirement: Lockout after 1..LOCKOUT_THRESHOLD_MAX failed attempts (0 = never locks). How to fix on Linux: (1) Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)). (2) Set deny = 5 in /etc/security/faillock.conf. How to verify: Re-run the audit; Account lockout threshold should be between 1 and 5. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Lockout observation period on macOS: Lockout observation period on macOS. Why it matters: Failed attempts must be counted over a long enough window to detect slow guessing. Risk if missing: Low - A short window lets attackers pace their guesses to avoid lockout. Requirement: Failed-attempt counter kept for >= LOCKOUT_OBSERVATION_MIN_MINUTES minutes. How to fix on macOS: (1) On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile. How to verify: Re-run the audit; Lockout observation period should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Lockout observation period on Linux: Lockout observation period on Linux. Why it matters: Failed attempts must be counted over a long enough window to detect slow guessing. Risk if missing: Low - A short window lets attackers pace their guesses to avoid lockout. Requirement: Failed-attempt counter kept for >= LOCKOUT_OBSERVATION_MIN_MINUTES minutes. How to fix on Linux: (1) Set fail_interval = 900 (seconds) in /etc/security/faillock.conf. How to verify: Re-run the audit; Lockout observation period should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Debian GNU/Linux 12 (bookworm) 12 (linux), audited 2026-10-02; 15 of 44 assessed checks compliant; 29 gaps, 5 pending, 0 not read. Gaps: Minimum password length: 0; Password expiration: PASS_MAX_DAYS 99999; Minimum password age: PASS_MIN_DAYS 0; Password complexity: minclass=2, required classes=1; Account lockout threshold: 10; Account lockout duration: no lockout; Lockout observation period: 10 minutes; ClamAV malware protection service: not running.\n\nQUESTION: How do I fix lockout observation period?"}, {"role": "assistant", "content": "Your audit found Lockout observation period: 10 minutes. Failed attempts must be counted over a long enough window to detect slow guessing. To fix it [4]:\n1. Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "linux", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup is non-compliant\", \"finding\": \"iCloud Backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store is non-compliant\", \"finding\": \"Apps only from the App Store is apps from outside the App Store instead of App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: macOS 26.0.1\nREQUIRED: macOS 14 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (macos):\n- Upgrade to macOS 14 or later, or replace Macs that cannot run a supported version.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Compliant\", \"title\": \"Supported operating system version is compliant\", \"finding\": \"Supported operating system version is set to macOS 26.0.1, in line with the required macOS 14 or later.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised is non-compliant\", \"finding\": \"Found: analytics sharing on. Required: analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000570, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"Found: no. Required: yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: jdoe\nREQUIRED: password required\nDETAIL: Users checked: jdoe\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"The device reports not required for: jdoe, but the baseline requires password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access: meets the baseline\", \"finding\": \"USB storage device access is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (windows):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"Found: no AI tools found. This satisfies the baseline requirement of inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Red Hat Enterprise Linux 8.10 (Ootpa)\nREQUIRED: RHEL 9 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"The device reports Red Hat Enterprise Linux 8.10 (Ootpa), but the baseline requires RHEL 9 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iPadOS 26.7 (ios)\nHOSTNAME: FIN-iPad-28\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=7, Non-Compliant=16, NotApplicable=0, Pending=6, Error=0\nEVIDENCE: 2 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Non-Compliant | Auto-Lock Never\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Non-Compliant | jailbroken\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store only\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iPadOS 26.7 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 15 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | download only\n- gdpr.os_supported | Supported operating system version | Compliant | iPadOS 26.7\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | DeepSeek, Meta AI\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | DeepSeek, Meta AI\n- aiact.risk_classification | AI system risk classification | Pending | DeepSeek, Meta AI\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | no\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- screen_lock.timeout: Medium\n- mobile.passcode_strength: Medium\n- mobile.jailbreak_root: High\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.ai_register: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=7, assessed=23, percent=30\nTOP GAPS (computed, highest risk first):\n- mobile.jailbreak_root | Not jailbroken | High | jailbroken\n- vuln.pending_updates | iOS security updates | High | iPadOS 26.7 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 15 critical/high\n- gdpr.auto_updates | Automatic iOS updates | High | download only\n- gdpr.backup_configured | iCloud Backup | High | no backup\nFIX GUIDANCE (highest-risk gaps):\n- mobile.jailbreak_root: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"FIN-iPad-28 - iPadOS 26.7\", \"score\": {\"compliant\": 7, \"assessed\": 23, \"percent\": 30}, \"counts\": {\"Compliant\": 7, \"Non-Compliant\": 16, \"NotApplicable\": 0, \"Pending\": 6, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.jailbreak_root\", \"name\": \"Not jailbroken\", \"risk\": \"High\", \"found\": \"jailbroken\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iPadOS 26.7 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"15 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"download only\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}], \"summary\": \"FIN-iPad-28 meets 7 of 23 assessed checks (30%). The most serious gaps are Not jailbroken, iOS security updates and Known vulnerabilities in iOS. 6 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Not jailbroken: Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 4 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked is non-compliant\", \"finding\": \"The user reports Always, but the baseline requires When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: DeepSeek, Grok\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"The user reports DeepSeek, Grok, but the baseline requires only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Found: allow (no firewall). Required: deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: analytics sharing off\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Compliant\", \"title\": \"Diagnostic data minimised is compliant\", \"finding\": \"Diagnostic data minimised is set to analytics sharing off, in line with the required analytics sharing off.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001351, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup is non-compliant\", \"finding\": \"Found: no backup. Required: backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamav-daemon\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Compliant\", \"title\": \"ClamAV malware protection service is compliant\", \"finding\": \"ClamAV malware protection service is set to clamav-daemon, in line with the required running.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates is non-compliant\", \"finding\": \"App updates is automatic app updates off instead of automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Claude\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unapproved AI tools (shadow AI)\", \"finding\": \"Found: ChatGPT, Claude. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 18.7 (ios)\nAI TOOLS FOUND: ChatGPT, Claude\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Claude.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: ClamAV malware protection service\", \"finding\": \"The device reports not running, but the baseline requires running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 1 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Compliant\", \"title\": \"Minimum password age: meets the baseline\", \"finding\": \"The device reports 1 days, which meets the requirement (>= 1 days).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 41 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 19 applications inventoried, 12 recognised (23 common applications covered). VLC 3.0.19 (2 CVEs, critical; fixed in 3.0.20; e.g. CVE-2023-47359, CVE-2023-47360); Adobe Acrobat Reader 26.001.21431 (36 CVEs, high; fixed in 26.002.21901; e.g. CVE-2026-47911, CVE-2026-47912); OpenVPN Connect 3.4.4 (3 CVEs, high; fixed in 3.5.0; e.g. CVE-2023-7224, CVE-2023-7245). 18 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 41 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 0 days\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\n- For a stand-alone device: net accounts /minpwage:1\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"The device reports 0 days, but the baseline requires >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Minimum password age = 1 day(s).\", \"For a stand-alone device: net accounts /minpwage:1\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Passcode strength\", \"finding\": \"Found: 4-digit code. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] What is in the audit report: The report has a compliance score, the most important gaps, and for each problem: what was found, why it matters, the risk level, step-by-step fixes for that operating system, how to verify the fix, and the related ISO 27001, GDPR and EU AI Act references. It also lists AI tools found and their EU AI Act risk class.\n[2] Privacy: where do audit results go?: Nowhere. Evidence and reports stay on the device. The app downloads only two things: the AI auditor model and a signed list of known vulnerabilities. Nothing about the device is uploaded. Chat history is stored only in this browser.\n[3] Automatic security updates on Linux: Automatic security updates on Linux. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on Linux: (1) Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer. How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Minimum password length on Linux: Minimum password length on Linux. Why it matters: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength. Risk if missing: Medium - Weak passwords make account takeover through guessing or credential stuffing easier. Requirement: Configured minimum length >= PASSWORD_MIN_LENGTH. How to fix on Linux: (1) Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack. (2) Set minlen = 14 in /etc/security/pwquality.conf. How to verify: Re-run the audit; Minimum password length should read 14 or more. Related controls: ISO 27001:2022 A.5.17 (Authentication information); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Debian GNU/Linux 12 (bookworm) 12 (linux), audited 2026-10-02; 30 of 47 assessed checks compliant; 17 gaps, 6 pending, 1 not read. Gaps: Minimum password length: 12; Account lockout duration: 10 minutes; ClamAV malware protection service: not running; On-access malware protection: not running; Firewall: no active firewall; Firewall inbound policy: allow (no firewall); Firewall outbound policy: no firewall; Screen lock timeout: too long for: jdoe (0s).\n\nQUESTION: What is the most important thing to fix?"}, {"role": "assistant", "content": "Your audit found 17 gaps. Start with Minimum password length (found: 12). Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength. To fix it [4]:\n1. Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n2. Set minlen = 14 in /etc/security/pwquality.conf.\n\nThen look at: Account lockout duration, ClamAV malware protection service, On-access malware protection."}], "meta": {"task": "chat", "os": "linux", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.7 checked against NVD iOS records. iOS 26.7 (15 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-43688). 32 medium/low.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS: does not meet the baseline\", \"finding\": \"The user reports 15 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_pwquality not enabled\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"The device reports pam_pwquality not enabled, but the baseline requires 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Human oversight for high-risk AI\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Gemini\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Unapproved AI tools (shadow AI) is Gemini instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (ios):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store is non-compliant\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAI TOOLS FOUND: Ollama\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /home/jdoe/.ollama/models (27G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Ollama.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_faillock not enabled\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (linux):\n- Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\n- Set deny = 5 in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold is non-compliant\", \"finding\": \"Account lockout threshold is pam_faillock not enabled instead of 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Enable pam_faillock in the PAM auth stack (RHEL: authselect enable-feature with-faillock; Debian/Ubuntu: add the preauth/authfail lines described in pam_faillock(8)).\", \"Set deny = 5 in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"The device reports allow (no firewall), but the baseline requires deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"The device reports USB storage writable without enforced encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: decision pending\", \"finding\": \"Found: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: Claude Code.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length is non-compliant\", \"finding\": \"The device reports 10, but the baseline requires >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Standard (mkhan)\nREQUIRED: Standard\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Compliant\", \"title\": \"Signed-in user account type: meets the baseline\", \"finding\": \"The device reports Standard (mkhan), which meets the requirement (Standard).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000942, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"Found: not set. Required: >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 4\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Compliant\", \"title\": \"Account lockout threshold: meets the baseline\", \"finding\": \"Found: 4. This satisfies the baseline requirement of 1-5.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: Defender in Passive Mode\nREQUIRED: on\nDETAIL: Another antivirus is primary: CrowdStrike Falcon\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (windows):\n- Enable real-time protection in Microsoft Defender (on-access protection follows it).\nverify: Re-run the audit; On-access malware protection should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"NotApplicable\", \"title\": \"On-access protection: not applicable\", \"finding\": \"On-access protection is not assessed here: Another antivirus is primary: CrowdStrike Falcon.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (windows):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"Human oversight for high-risk AI is partial instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_BREACH_PROCEDURE in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect 5388\nREQUIRED: antivirus present\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (macos):\n- XProtect is built in; if it is missing or damaged, reinstall the latest macOS update. Add a managed endpoint protection product if your policy requires on-access scanning.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus matches the required setting\", \"finding\": \"Antivirus is set to XProtect 5388, in line with the required antivirus present.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted: does not meet the baseline\", \"finding\": \"Apple Intelligence external integrations restricted is ChatGPT extension on instead of external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: rdp\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Remote access services restricted is rdp instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAI TOOLS FOUND: GPT4All\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- GPT4All: Local model runner; provider Nomic AI (runs locally); data: On the device unless optional data sharing is enabled; default risk minimal; GDPR: Keep the optional data lake sharing switched off.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"GPT4All\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Keep the optional data lake sharing switched off.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: GPT4All.\"}"}], "meta": {"task": "ai_classification", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: linux-image-generic, openssh-server, libsystemd0. Package lists are 30 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Pending security updates is non-compliant\", \"finding\": \"Found: 4 pending. Required: <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; no signed-in user to check\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"Found: no machine inactivity limit; no signed-in user to check. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 26.5.2 (macos)\nAI TOOLS FOUND: Cursor, Ollama, Claude\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: Local models: /Users/jdoe/.ollama/models (33G).\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Cursor: AI code editor; provider Anysphere (US); data: Code sent to the provider and its model providers unless privacy mode is enforced; default risk minimal; GDPR: Enforce privacy mode through a business plan and sign a data processing agreement.\n- Ollama: Local model runner; provider Open source (runs locally); data: On the device; nothing leaves unless a cloud model is selected; default risk minimal; GDPR: No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Cursor\", \"kind\": \"AI code editor\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Enforce privacy mode through a business plan and sign a data processing agreement.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Ollama\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No transfer to a provider, but check each model's licence and keep the local API (port 11434) closed to the network.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Cursor, Ollama, Claude.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Private outbound is non-compliant\", \"finding\": \"The device reports allow, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Auto-Lock: does not meet the baseline\", \"finding\": \"Found: Auto-Lock Never. Required: <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 57 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-02 Cumulative Update for Windows 11 (2026-08-10)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently is non-compliant\", \"finding\": \"The device reports 57 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"Found: partial. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training: meets the baseline\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rooted or bootloader unlocked\nREQUIRED: not rooted\nDETAIL: Android security protections are weakened on a rooted phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not rooted is non-compliant\", \"finding\": \"The user reports rooted or bootloader unlocked, but the baseline requires not rooted. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"The device reports none found, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (macos):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not applicable\", \"finding\": \"Personal data discovery is not assessed here: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: external AI off unless approved\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Pending\", \"title\": \"Apple Intelligence external integrations restricted is pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Human oversight for high-risk AI is no instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security intelligence [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 19 days old (version 1.440.646.0)\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (windows):\n- Run Update-MpSignature and check that the device can reach Windows Update or your update server.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Security intelligence is non-compliant\", \"finding\": \"Found: 19 days old (version 1.440.646.0). Required: <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"Run Update-MpSignature and check that the device can reach Windows Update or your update server.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (CONTOSO\\pmartin)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"Found: Administrator (CONTOSO\\\\pmartin). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Install unknown apps blocked\", \"finding\": \"The user reports unknown apps allowed for at least one app, but the baseline requires all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"Found: none of the approved accounts is an administrator. Required: itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised: does not meet the baseline\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 38 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: systemd, libxml2, linux-image-generic. Package lists are 3 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"Pending security updates is 38 pending instead of <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: minclass=0, required classes=0\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Found: minclass=0, required classes=0. Required: 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: decision pending\", \"finding\": \"Answer ORG_ROPA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (macos):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answer ORG_AI_TRANSPARENCY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after 60 seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"The device reports password after 60 seconds, but the baseline requires password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Compliant\", \"title\": \"Public outbound matches the required setting\", \"finding\": \"Public outbound is set to allow, in line with the required allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private firewall [firewall.private.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Private firewall: does not meet the baseline\", \"finding\": \"Private firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (windows):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (jdoe)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"The device reports Administrator (jdoe), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000386, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none found\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (windows):\n- Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Backup configured is none found instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Back up user folders with OneDrive Known Folder Move (Group Policy/Intune), File History, or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 57 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-04 Cumulative Update for Windows 11 (2026-08-06)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently is non-compliant\", \"finding\": \"Found: 57 days ago. Required: <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Firewall inbound policy\", \"finding\": \"Found: allow (no firewall). Required: deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Answer ORG_DPIA in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"Firewall inbound policy is firewall off, stealth mode off instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration is non-compliant\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\", \"For a stand-alone device: net accounts /maxpwage:365\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_pwquality not enabled\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity is non-compliant\", \"finding\": \"Found: pam_pwquality not enabled. Required: 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Found: ChatGPT. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"Found: partial. Required: yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000743, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Privacy: where do audit results go?: Nowhere. Evidence and reports stay on the device. The app downloads only two things: the AI auditor model and a signed list of known vulnerabilities. Nothing about the device is uploaded. Chat history is stored only in this browser.\n[2] Automatic security updates on Linux: Automatic security updates on Linux. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on Linux: (1) Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer. How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Password history on Linux: Password history on Linux. Why it matters: Without a password history, users can return to old, possibly exposed passwords. Risk if missing: Low - Reused passwords keep compromised credentials valid. Requirement: Remembered passwords >= PASSWORD_HISTORY. How to fix on Linux: (1) Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok'). (2) Or set remember = 24 in /etc/security/pwhistory.conf where supported. How to verify: Re-run the audit; Password history should read 24 or more. Related controls: ISO 27001:2022 A.5.17 (Authentication information); ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] What is in the audit report: The report has a compliance score, the most important gaps, and for each problem: what was found, why it matters, the risk level, step-by-step fixes for that operating system, how to verify the fix, and the related ISO 27001, GDPR and EU AI Act references. It also lists AI tools found and their EU AI Act risk class.\n\nMY AUDIT: Ubuntu 24.04.3 LTS 24.04 (linux), audited 2026-10-03; 32 of 46 assessed checks compliant; 14 gaps, 7 pending, 1 not read. Gaps: Password history: 10; Password expiration: PASS_MAX_DAYS 90; accounts: jdoe (99999); Account lockout duration: 10 minutes; Antivirus: none running; Malware signatures: no signature database; On-access malware protection: not running; Screen lock timeout: too long for: asmith (3600s); USB storage write: allow.\n\nQUESTION: What is the most important thing to fix?"}, {"role": "assistant", "content": "Your audit found 14 gaps. Start with Password history (found: 10). Without a password history, users can return to old, possibly exposed passwords. To fix it [3]:\n1. Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n2. Or set remember = 24 in /etc/security/pwhistory.conf where supported.\n\nThen look at: Password expiration, Account lockout duration, Antivirus."}], "meta": {"task": "chat", "os": "linux", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities is non-compliant\", \"finding\": \"Record of processing activities is no instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI assistant access restricted\", \"finding\": \"Found: assistant screen access on. Required: screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"The device reports partial, but the baseline requires yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 14\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Compliant\", \"title\": \"Minimum password length matches the required setting\", \"finding\": \"Found: 14. This satisfies the baseline requirement of >= 14.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted: does not meet the baseline\", \"finding\": \"The user reports assistant screen access on, but the baseline requires screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates: does not meet the baseline\", \"finding\": \"Automatic security updates is check=1 download=1 security=1 macOS-install=0 instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.7 (ios)\nHOSTNAME: DEV-iPhone-05\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=15, Non-Compliant=4, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 4 minutes\n- mobile.passcode_strength | Passcode strength | Compliant | 7+ digit numeric code\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Compliant | App Store only\n- mobile.find_my_device | Find My iPhone | Compliant | on\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.7 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 15 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.7\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Compliant | iCloud Backup on\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Non-Compliant | Always\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Microsoft Copilot\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Microsoft Copilot\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Microsoft Copilot\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | ChatGPT extension off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.lock_screen_previews: Low\n- aiact.unapproved: Medium\nSCORE (computed): compliant=15, assessed=19, percent=79\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | iOS security updates | High | iOS 26.7 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 15 critical/high\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | ChatGPT, Microsoft Copilot\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Low | Always\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it."}, {"role": "assistant", "content": "{\"device\": \"DEV-iPhone-05 - iOS 26.7\", \"score\": {\"compliant\": 15, \"assessed\": 19, \"percent\": 79}, \"counts\": {\"Compliant\": 15, \"Non-Compliant\": 4, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.7 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"15 critical/high\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"ChatGPT, Microsoft Copilot\"}, {\"check_id\": \"gdpr.lock_screen_previews\", \"name\": \"Notification previews hidden when locked\", \"risk\": \"Low\", \"found\": \"Always\"}], \"summary\": \"DEV-iPhone-05 meets 15 of 19 assessed checks (79%). The most serious gaps are iOS security updates and Known vulnerabilities in iOS. 10 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Found: none. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Automatic security updates\", \"finding\": \"The device reports check=1 download=1 security=1 macOS-install=0, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAI TOOLS FOUND: Claude Code, ChatGPT, Microsoft Copilot\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Microsoft Copilot: Cloud AI assistant; provider Microsoft (US); data: Microsoft cloud; enterprise plans keep data under Microsoft's commercial data protection terms; default risk minimal; GDPR: Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Microsoft Copilot\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude Code, ChatGPT, Microsoft Copilot.\"}"}], "meta": {"task": "ai_classification", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: writing to USB storage blocked\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Compliant\", \"title\": \"Removable media encryption matches the required setting\", \"finding\": \"Found: writing to USB storage blocked. This satisfies the baseline requirement of no unencrypted writes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: machine inactivity limit locks the session\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement: meets the baseline\", \"finding\": \"Screen lock password requirement is set to machine inactivity limit locks the session, in line with the required sign-in required to unlock.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"The device reports none of the approved accounts is an administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: root filesystem not encrypted\nREQUIRED: system disk encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (linux):\n- Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"Encryption is root filesystem not encrypted instead of system disk encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI: does not meet the baseline\", \"finding\": \"Human oversight for high-risk AI is no instead of yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 59 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 58; flatpaks: us.zoom.Zoom\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"Found: 59 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: screen lock set\nREQUIRED: PIN, password or pattern\nDETAIL: The browser reports that a device unlock method (passcode, PIN, Face ID or fingerprint) is set up.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose PIN or Password.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Screen lock is compliant\", \"finding\": \"The device reports screen lock set, which meets the requirement (PIN, password or pattern).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000043, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"Approved IT administrator is none of the approved accounts is an administrator instead of itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: Play Protect scanning on\nREQUIRED: on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Google Play Protect is compliant\", \"finding\": \"Google Play Protect is set to Play Protect scanning on, in line with the required on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (windows):\n- To block USB storage completely: Group Policy > Administrative Templates > System > Removable Storage Access > All Removable Storage classes: Deny all access = Enabled, or set the USBSTOR service Start value to 4.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access is compliant\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"Found: none. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (rschmit)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"Found: Administrator (rschmit). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: set automatically\nREQUIRED: set automatically\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Date & time set automatically matches the required setting\", \"finding\": \"Found: set automatically. This satisfies the baseline requirement of set automatically. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 10 (android)\nHOSTNAME: ops-galaxy-32\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=6, Non-Compliant=13, NotApplicable=2, Pending=9, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Non-Compliant | Play Protect scanning off\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 2 minutes\n- mobile.passcode_strength | Screen lock strength | Non-Compliant | pattern\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Non-Compliant | unknown apps allowed for at least one app\n- mobile.usb_debugging | USB debugging off | Non-Compliant | on\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 10 encrypts by default)\n- vuln.pending_updates | Android security update | Non-Compliant | security patch 2025-02 (20 months old)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- gdpr.auto_updates | Automatic system updates | Non-Compliant | manual\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Android 10\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | Google backup | Non-Compliant | no backup\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Pending | not answered\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Non-Compliant | all content shown\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | AI assistant access restricted | Non-Compliant | assistant screen access on\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- malware.antivirus: High\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.usb_debugging: Medium\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- gdpr.auto_updates: High\n- gdpr.os_supported: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.lock_screen_previews: Low\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=6, assessed=19, percent=32\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Google Play Protect | High | Play Protect scanning off\n- mobile.app_sources | Install unknown apps blocked | High | unknown apps allowed for at least one app\n- vuln.pending_updates | Android security update | High | security patch 2025-02 (20 months old)\n- gdpr.auto_updates | Automatic system updates | High | manual\n- gdpr.os_supported | Supported operating system version | High | Android 10\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\n- mobile.app_sources: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\n- vuln.pending_updates: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone."}, {"role": "assistant", "content": "{\"device\": \"ops-galaxy-32 - Android 10\", \"score\": {\"compliant\": 6, \"assessed\": 19, \"percent\": 32}, \"counts\": {\"Compliant\": 6, \"Non-Compliant\": 13, \"NotApplicable\": 2, \"Pending\": 9, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Google Play Protect\", \"risk\": \"High\", \"found\": \"Play Protect scanning off\"}, {\"check_id\": \"mobile.app_sources\", \"name\": \"Install unknown apps blocked\", \"risk\": \"High\", \"found\": \"unknown apps allowed for at least one app\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Android security update\", \"risk\": \"High\", \"found\": \"security patch 2025-02 (20 months old)\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic system updates\", \"risk\": \"High\", \"found\": \"manual\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Android 10\"}], \"summary\": \"ops-galaxy-32 meets 6 of 19 assessed checks (32%). The most serious gaps are Google Play Protect, Install unknown apps blocked and Android security update. 9 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Google Play Protect: Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\", \"Install unknown apps blocked: Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\", \"Android security update: Settings › System › Software updates (and Security & privacy › Updates): install the latest system and Google Play system updates. If the security patch stays older than 90 days because the maker no longer sends updates, replace the phone.\", \"Answer the 8 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification does not apply here\", \"finding\": \"Not applicable on this device: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"AI transparency to users is partial instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: Veeam Agent for Microsoft Windows.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Confirm the last backup date in: Veeam Agent for Microsoft Windows. Current state: date not readable.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nDETAIL: macOS resets the failed-attempt count when the lockout ends.\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (macos):\n- On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Lockout observation period is non-compliant\", \"finding\": \"Found: not set. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nHOSTNAME: LAPTOP-1WZWAMC\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=36, Non-Compliant=21, NotApplicable=3, Pending=3, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 60 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | enabled\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Error | not read\n- lockout.observation | Lockout observation period | Non-Compliant | no lockout configured\n- malware.antivirus | Antivirus | Non-Compliant | Windows Defender (off, out of date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Non-Compliant | off\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Compliant | on\n- malware.signatures | Security intelligence | Compliant | 3 days old (version 1.432.452.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Compliant | allow\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Compliant | on\n- firewall.public.inbound | Public inbound | Compliant | block\n- firewall.public.outbound | Public outbound | Non-Compliant | block\n- screen_lock.enabled | Screen lock | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | CONTOSO\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (CONTOSO\\lmuller)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Compliant | protection On, XtsAes128\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 2 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 50 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | Windows default (automatic)\n- gdpr.updates_recent | Security updates installed recently | Compliant | 3 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Microsoft Windows 10 Pro build 19045\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NTP, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | ssh, teamviewer\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | not configured\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Non-Compliant | Recall feature Enabled, not disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- lockout.observation: Low\n- malware.antivirus: High\n- malware.realtime: High\n- firewall.domain.inbound: High\n- firewall.private.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.os_ai_features: Medium\n- org.processor_agreements: High\n- org.human_oversight: Medium\nSCORE (computed): compliant=36, assessed=57, percent=63\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | Windows Defender (off, out of date)\n- malware.realtime | Real-time protection | High | off\n- firewall.domain.inbound | Domain inbound | High | allow\n- firewall.private.inbound | Private inbound | High | allow\n- user.account_type | Signed-in user account type | High | Administrator (CONTOSO\\lmuller)\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\n- malware.realtime: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block."}, {"role": "assistant", "content": "{\"device\": \"LAPTOP-1WZWAMC - Microsoft Windows 10 Pro 10.0.19045\", \"score\": {\"compliant\": 36, \"assessed\": 57, \"percent\": 63}, \"counts\": {\"Compliant\": 36, \"Non-Compliant\": 21, \"NotApplicable\": 3, \"Pending\": 3, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"Windows Defender (off, out of date)\"}, {\"check_id\": \"malware.realtime\", \"name\": \"Real-time protection\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (CONTOSO\\\\lmuller)\"}], \"summary\": \"LAPTOP-1WZWAMC meets 36 of 57 assessed checks (63%). The most serious gaps are Antivirus, Real-time protection and Domain inbound. 1 checks could not be read. 3 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Antivirus: Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center.\", \"Real-time protection: Set-MpPreference -DisableRealtimeMonitoring $false, and remove any policy that disables real-time protection.\", \"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Answer the 3 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: password after 0 seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Compliant\", \"title\": \"Screen lock password requirement: meets the baseline\", \"finding\": \"The device reports password after 0 seconds, which meets the requirement (password within 5 seconds).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 15 (android)\nAI TOOLS FOUND: ChatGPT\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Hugging Face models, Ollama, Claude Code\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: Hugging Face models, Ollama, Claude Code. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000510, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 89 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 749 packages checked against rhel:9 advisories. samba-devel 0:4.16.4-101.el9 (14 CVEs, critical; fixed in 0:4.23.5-10.el9_8; e.g. CVE-2022-2127, CVE-2023-34966); ipa-common 0:4.10.2-5.el9_3 (12 CVEs, critical; fixed in 0:4.13.4-1.el9_8; e.g. CVE-2024-2698, CVE-2024-3183); rubygem-rss 0:0.3.1-7.module+el9.8.0+24406+8975e57f (12 CVEs, critical; fixed in 0:0.3.2-35.module+el9.8.0+24852+4534d062; e.g. CVE-2026-27820, CVE-2026-42256); samba-winbind-modules-debuginfo 0:4.18.6-100.el9 (9 CVEs, critical; fixed in 0:4.23.5-10.el9_8; e.g. CVE-2023-3961, CVE-2023-4091); python3-net-snmp-debuginfo 1:5.9.1-9.el9 (1 CVE, critical; fixed in 1:5.9.1-17.el9_7.1; e.g. CVE-2025-68615); and 9 more. 25 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"The device reports 89 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAI TOOLS FOUND: Jan\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Jan: Local model runner; provider Open source (runs locally); data: On the device unless a cloud model is connected; default risk minimal; GDPR: Check whether remote API providers are configured; if so, the cloud provider rules apply.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Jan\", \"kind\": \"Local model runner\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Check whether remote API providers are configured; if so, the cloud provider rules apply.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Jan.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: winrm\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Remote access services restricted\", \"finding\": \"Remote access services restricted is winrm instead of only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: screen saver does not require sign-in for: DESKTOP-R2S2I1D\\mkhan\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"Found: screen saver does not require sign-in for: DESKTOP-R2S2I1D\\\\mkhan. Required: sign-in required to unlock. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 4\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Compliant\", \"title\": \"Account lockout threshold is compliant\", \"finding\": \"Found: 4. This satisfies the baseline requirement of 1-5.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 27.0 (ios)\nHOSTNAME: MKT-iPhone-07\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=14, Non-Compliant=8, NotApplicable=3, Pending=4, Error=0\nEVIDENCE: 5 detected by the browser, 16 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 2 minutes\n- mobile.passcode_strength | Passcode strength | Non-Compliant | 4-digit code\n- mobile.jailbreak_root | Not jailbroken | Pending | not answered\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Compliant | iOS 27.0 (up to date)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Non-Compliant | off\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 27.0\n- gdpr.time_sync | Date & time set automatically | Compliant | set automatically\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Compliant | analytics sharing off\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | NotApplicable | Apple Intelligence not available\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.passcode_strength: Medium\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- gdpr.auto_updates: High\n- gdpr.backup_configured: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.ai_register: Medium\nSCORE (computed): compliant=14, assessed=22, percent=64\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- gdpr.auto_updates | Automatic iOS updates | High | off\n- gdpr.backup_configured | iCloud Backup | High | no backup\n- org.breach_procedure | Breach response procedure | High | partial\n- mobile.passcode_strength | Passcode strength | Medium | 4-digit code\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- gdpr.auto_updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\n- gdpr.backup_configured: Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps."}, {"role": "assistant", "content": "{\"device\": \"MKT-iPhone-07 - iOS 27.0\", \"score\": {\"compliant\": 14, \"assessed\": 22, \"percent\": 64}, \"counts\": {\"Compliant\": 14, \"Non-Compliant\": 8, \"NotApplicable\": 3, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic iOS updates\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"org.breach_procedure\", \"name\": \"Breach response procedure\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"mobile.passcode_strength\", \"name\": \"Passcode strength\", \"risk\": \"Medium\", \"found\": \"4-digit code\"}], \"summary\": \"MKT-iPhone-07 meets 14 of 22 assessed checks (64%). The most serious gaps are Apps only from the App Store, Automatic iOS updates and iCloud Backup. 4 items are pending. 16 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"Automatic iOS updates: Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\", \"iCloud Backup: Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\", \"Answer the 3 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound: does not meet the baseline\", \"finding\": \"Private inbound is allow instead of block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates: does not meet the baseline\", \"finding\": \"App updates is automatic app updates off instead of automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (FABRIKAM\\jdoe)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Signed-in user account type is Administrator (FABRIKAM\\\\jdoe) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: macOS 15.6 (macos)\nAI TOOLS FOUND: Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 314 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 16 applications inventoried, 13 recognised (31 common applications covered). Google Chrome 152.0.7977.65 (282 CVEs, critical, actively exploited; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301); Microsoft Visual Studio Code (User) 1.109.2 (25 CVEs, critical; fixed in 1.136.2; e.g. CVE-2026-40376, CVE-2026-41109); Microsoft Teams 1.6.00.26474 (6 CVEs, high; fixed in 25163.3001.3726.6503; e.g. CVE-2023-29328, CVE-2023-29330); LibreOffice 24.2.5 (1 CVE, high; fixed in 24.2.5.1; e.g. CVE-2024-6472). 337 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in installed software\", \"finding\": \"Found: 314 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Personal data discovery [gdpr.personal_data_discovery] in GDPR TECHNICAL CONTROLS\nSTATUS: NotApplicable\nFOUND: not run\nREQUIRED: opt-in scan\nDETAIL: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- ISO 27001:2022 A.5.12 Classification of information: Label information by how sensitive it is so that the right protection can be applied.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 5(1)(e) Storage limitation: Personal data must not be kept in identifiable form longer than necessary.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: Knowing where personal data is stored on endpoints is needed for the record of processing and for storage limitation.\nrisk: Low - Not assessed in stage 1.\nfix (macos):\n- No action needed now: this opt-in scan is planned for a later stage of the tool.\nverify: Not applicable in stage 1.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.personal_data_discovery\", \"status\": \"NotApplicable\", \"title\": \"Personal data discovery: not assessed on this device\", \"finding\": \"Personal data discovery is not assessed here: Opt-in scan (PERSONAL_DATA_SCAN=0) is planned for a later stage.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"ISO 27001:2022 A.5.12\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 5(1)(e)\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.personal_data_discovery", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1200 seconds (machine inactivity limit)\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock timeout: does not meet the baseline\", \"finding\": \"The device reports 1200 seconds (machine inactivity limit), but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAI TOOLS FOUND: Claude, Microsoft Copilot\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude: Cloud AI assistant; provider Anthropic (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\n- Microsoft Copilot: Cloud AI assistant; provider Microsoft (US); data: Microsoft cloud; enterprise plans keep data under Microsoft's commercial data protection terms; default risk minimal; GDPR: Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a commercial plan with a data processing agreement and a transfer basis (Art. 28, Art. 44).\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Microsoft Copilot\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude, Microsoft Copilot.\"}"}], "meta": {"task": "ai_classification", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic system updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: manual\nREQUIRED: automatic\nDETAIL: Security updates wait for the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (android):\n- Settings › System › Software updates: turn on automatic download (names vary by phone maker).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic system updates: does not meet the baseline\", \"finding\": \"Found: manual. Required: automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › System › Software updates: turn on automatic download (names vary by phone maker).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"Found: no. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 60 seconds\nREQUIRED: <= 300 seconds\nDETAIL: Idle 60s (screen saver/display sleep) + 0s password delay.\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (macos):\n- Set the screen saver to start within 5 minutes (idleTime in com.apple.screensaver) and require the password immediately, ideally by configuration profile.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout matches the required setting\", \"finding\": \"The device reports 60 seconds, which meets the requirement (<= 300 seconds).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 11 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 11\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"Found: 11 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: Veeam Agent for Microsoft Windows.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Found: date not readable. Confirm the last backup date in: Veeam Agent for Microsoft Windows.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: App updates\", \"finding\": \"Found: automatic app updates off. Required: automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates: does not meet the baseline\", \"finding\": \"Found: automatic app updates off. Required: automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off for: lmuller\nREQUIRED: on\nDETAIL: Users checked: lmuller\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (linux):\n- For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock: does not meet the baseline\", \"finding\": \"The device reports off for: lmuller, but the baseline requires on. An idle device should lock itself so nobody can use it while the owner is away.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"An unlocked, unattended device exposes all the user's data and accounts.\"}, \"recommendation\": [\"For GNOME: gsettings set org.gnome.desktop.screensaver lock-enabled true, and enforce it for all users with a locked dconf setting in /etc/dconf/db/local.d/.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: backup on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Pending\", \"title\": \"iCloud Backup is pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound: does not meet the baseline\", \"finding\": \"Public outbound is allow instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Sensitive notifications hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: all content shown\nREQUIRED: sensitive content hidden\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (android):\n- Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Sensitive notifications hidden when locked\", \"finding\": \"Sensitive notifications hidden when locked is all content shown instead of sensitive content hidden. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Notifications on lock screen: choose Hide sensitive content or Don't show notifications.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"Found: no lockout. Required: >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Account lockout duration on Windows: Account lockout duration on Windows. Why it matters: The lockout must last long enough to slow down repeated guessing. Risk if missing: Low - A short lockout lets attackers resume guessing almost at once. Requirement: Locked for >= LOCKOUT_DURATION_MIN_MINUTES minutes, or until an administrator unlocks. How to fix on Windows: (1) Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more. (2) For a stand-alone device: net accounts /lockoutduration:15 How to verify: Re-run the audit; Account lockout duration should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Lockout observation period on Windows: Lockout observation period on Windows. Why it matters: Failed attempts must be counted over a long enough window to detect slow guessing. Risk if missing: Low - A short window lets attackers pace their guesses to avoid lockout. Requirement: Failed-attempt counter kept for >= LOCKOUT_OBSERVATION_MIN_MINUTES minutes. How to fix on Windows: (1) Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more. (2) For a stand-alone device: net accounts /lockoutwindow:15 How to verify: Re-run the audit; Lockout observation period should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Account lockout threshold on macOS: Account lockout threshold on macOS. Why it matters: Locking an account after a few wrong passwords stops online guessing. Risk if missing: Medium - Unlimited attempts allow brute-force attacks against local accounts. Requirement: Lockout after 1..LOCKOUT_THRESHOLD_MAX failed attempts (0 = never locks). How to fix on macOS: (1) Deploy a Passcode configuration profile with maxFailedAttempts = 5. How to verify: Re-run the audit; Account lockout threshold should be between 1 and 5. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Account lockout threshold on Windows: Account lockout threshold on Windows. Why it matters: Locking an account after a few wrong passwords stops online guessing. Risk if missing: Medium - Unlimited attempts allow brute-force attacks against local accounts. Requirement: Lockout after 1..LOCKOUT_THRESHOLD_MAX failed attempts (0 = never locks). How to fix on Windows: (1) Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0). (2) For a stand-alone device: net accounts /lockoutthreshold:5 How to verify: Re-run the audit; Account lockout threshold should be between 1 and 5. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: Microsoft Windows 11 Pro 10.0.22621 (windows), audited 2026-10-06; 12 of 53 assessed checks compliant; 41 gaps, 6 pending, 3 not read. Gaps: Minimum password length: 10; Minimum password age: 0 days; Password complexity: disabled; Account lockout threshold: 0; Account lockout duration: no lockout configured; Lockout observation period: 5 minutes; Antivirus: Windows Defender (off, out of date); On-access protection: off.\n\nQUESTION: How do I fix account lockout threshold?"}, {"role": "assistant", "content": "Your audit found Account lockout threshold: 0. Locking an account after a few wrong passwords stops online guessing. To fix it [4]:\n1. Set Group Policy: Account Policies > Account Lockout Policy > Account lockout threshold = 5 invalid attempts (or fewer, not 0).\n2. For a stand-alone device: net accounts /lockoutthreshold:5\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "windows", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (ios):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption is non-compliant\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"Found: none. Required: backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: DeepSeek, Gemini, Replika\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: DeepSeek, Gemini, Replika. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: >= 6 months for high-risk AI\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"NotApplicable\", \"title\": \"AI usage log retention does not apply here\", \"finding\": \"AI usage log retention is not assessed here: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 60 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 59; flatpaks: us.zoom.Zoom\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"Found: 60 outdated. Required: <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: logon: none; account management: none\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Non-Compliant\", \"title\": \"Security event logging is non-compliant\", \"finding\": \"Security event logging is logon: none; account management: none instead of logon success+failure and account changes audited. Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\"}, \"recommendation\": [\"Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Security event logging should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: on\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Compliant\", \"title\": \"Domain firewall: meets the baseline\", \"finding\": \"The device reports on, which meets the requirement (on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"Backup configured is none instead of backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Google Play Protect is non-compliant\", \"finding\": \"Found: Play Protect scanning off. Required: on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"The device reports no backup, but the baseline requires <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Error\nFOUND: not read\nREQUIRED: build 22631 or later\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (windows):\n- Upgrade to a supported Windows 11 release (build 22631 or later), or replace hardware that cannot run it.\nverify: Re-run the audit; Supported operating system version should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Error\", \"title\": \"Supported operating system version: evidence missing\", \"finding\": \"No evidence was collected for this check (Value could not be read.); its status is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (macos):\n- Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"Found: none of the approved accounts is an administrator. Required: itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 220 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-03 Cumulative Update for Windows 11 (2026-02-25)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Security updates installed recently is 220 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password length\", \"finding\": \"Minimum password length is 10 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store is non-compliant\", \"finding\": \"The user reports apps from outside the App Store, but the baseline requires App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: >= 6 months for high-risk AI\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"NotApplicable\", \"title\": \"AI usage log retention does not apply here\", \"finding\": \"This check does not apply: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 251 source packages checked against ubuntu:26.04 advisories. openssl 3.5.5-1ubuntu3.2 (1 CVE, high; fixed in 3.5.5-1ubuntu3.6; e.g. CVE-2026-84782). 47 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (linux):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 1 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the package manager to reach a fixed version (apt / dnf / snap / flatpak).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Compliant\", \"title\": \"AI register is compliant\", \"finding\": \"The device reports yes, which meets the requirement (yes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001173, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 269 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 21 applications inventoried, 10 recognised (23 common applications covered). Google Chrome 152.0.7977.75 (269 CVEs, critical, actively exploited; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301). 309 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 269 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 11 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: automatic app updates on\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Pending\", \"title\": \"App updates: decision pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001011, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (macos):\n- Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"Account lockout duration is not set instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with minutesUntilFailedLoginReset = 15.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised: does not meet the baseline\", \"finding\": \"The user reports usage & diagnostics on, but the baseline requires usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamav-daemon\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Compliant\", \"title\": \"ClamAV malware protection service: meets the baseline\", \"finding\": \"The device reports clamav-daemon, which meets the requirement (running).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000318, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Found: partial. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (windows):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Current state: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iCloud Backup\", \"finding\": \"The user reports no backup, but the baseline requires backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Found: no backup. Required: <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"The device reports none running, but the baseline requires antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 38 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 21 applications inventoried, 12 recognised (23 common applications covered). Adobe Acrobat Reader 25.001.21288 (38 CVEs, high, actively exploited; fixed in 26.002.21901; e.g. CVE-2026-34621, CVE-2026-34622). 19 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (macos):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software is non-compliant\", \"finding\": \"Known vulnerabilities in installed software is 38 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use the App Store, the vendor's updater or brew upgrade to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000715, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Password expiration: does not meet the baseline\", \"finding\": \"The device reports never, but the baseline requires 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: LM Studio\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Found: LM Studio. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: Defender in Passive Mode\nREQUIRED: on\nDETAIL: Another antivirus is primary: Sophos Endpoint\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (windows):\n- Enable real-time protection in Microsoft Defender (on-access protection follows it).\nverify: Re-run the audit; On-access malware protection should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"NotApplicable\", \"title\": \"On-access protection: not assessed on this device\", \"finding\": \"Not applicable on this device: Another antivirus is primary: Sophos Endpoint.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no active firewall\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Firewall is non-compliant\", \"finding\": \"The device reports no active firewall, but the baseline requires active. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security event logging [gdpr.security_logging] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: logon: success+failure; account management: success+failure\nREQUIRED: logon success+failure and account changes audited\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Logon and account-change events are needed to detect and investigate a breach within the 72-hour GDPR deadline.\nrisk: Medium - Without these logs, a breach cannot be scoped and the notification decision becomes guesswork.\nfix (windows):\n- Enable auditing: auditpol /set /subcategory:{0CCE9215-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (Logon) and auditpol /set /subcategory:{0CCE9235-69AE-11D9-BED3-505054503030} /success:enable /failure:enable (User Account Management), or set the Advanced Audit Policy by Group Policy.\nverify: Re-run the audit; Security event logging should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.security_logging\", \"status\": \"Compliant\", \"title\": \"Security event logging matches the required setting\", \"finding\": \"Security event logging is set to logon: success+failure; account management: success+failure, in line with the required logon success+failure and account changes audited.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"GDPR Art. 32(1)(d)\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.security_logging", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] AI software inventory on Android: AI software inventory on Android. Why it matters: Deployers must know which AI systems they use before they can classify and govern them. Risk if missing: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools. Requirement: AI apps, local AI services and local model folders are listed (Compliant = inventory taken). How to fix on Android: (1) Add the AI tools found on this device to the organisation's AI register, with owner and purpose. How to verify: The AI register lists every tool in the inventory. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); GDPR Art. 30 (Records of processing activities); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n[2] AI system risk classification on Android: AI system risk classification on Android. Why it matters: Each AI system must be checked against the banned practices and the high-risk uses in Annex III. Risk if missing: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards. Requirement: Each AI tool found is classified (prohibited / high-risk / limited / minimal). Done by the auditor model: Pending in the collector output, NotApplicable when no AI tool is found. How to fix on Android: (1) Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose. How to verify: Every AI tool in the inventory has a classification in the AI register. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); EU AI Act Art. 5 (Prohibited AI practices); EU AI Act Art. 6 (Classification rules for high-risk AI systems); EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)).\n[3] AI usage log retention on Android: AI usage log retention on Android. Why it matters: Deployers of high-risk AI systems must keep the system's logs for at least six months. Risk if missing: Medium - Missing logs prevent monitoring and investigation of high-risk AI use. Requirement: Logs kept >= 6 months if a high-risk AI system is used. Pending until classification; NotApplicable when no AI tool is found. How to fix on Android: (1) If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months. How to verify: Log retention is documented for each high-risk AI system. Related controls: ISO 27001:2022 A.8.15 (Logging); EU AI Act Art. 26(6) (Deployers: keep automatically generated logs for at least six months).\n[4] Unapproved AI tools (shadow AI) on Android: Unapproved AI tools (shadow AI) on Android. Why it matters: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Risk if missing: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations. Requirement: Every AI tool found is listed in APPROVED_AI_TOOLS. How to fix on Android: (1) Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it. (2) Publish an AI acceptable-use policy so staff know which tools are allowed. How to verify: Re-run the audit; Unapproved AI tools should read none. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 28 (Processor); GDPR Art. 44 (General principle for transfers); EU AI Act Art. 26 (Obligations of deployers of high-risk AI systems).\n\nMY AUDIT: Android 16 (android), audited 2026-10-06; 14 of 19 assessed checks compliant; 5 gaps, 11 pending, 0 not read. Gaps: Screen lock strength: 4-digit PIN; Automatic system updates: manual; Google backup: no backup; Sensitive notifications hidden when locked: all content shown; Unapproved AI tools (shadow AI): Character.AI.\n\nQUESTION: How do I fix unapproved ai tools (shadow ai)?"}, {"role": "assistant", "content": "Your audit found Unapproved AI tools (shadow AI): Character.AI. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. To fix it [4]:\n1. Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n2. Publish an AI acceptable-use policy so staff know which tools are allowed.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "android", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Where to find Install unknown apps blocked on Android: Install unknown apps blocked on Android: Is any app allowed to install unknown apps? Where to look: Settings › Apps › Special app access › Install unknown apps. Every app should show 'Not allowed'.\n[2] Install unknown apps blocked on Android: Install unknown apps blocked on Android. Why it matters: Official app stores review apps for malware; apps from other sources skip those checks. Risk if missing: High - Side-loaded apps are a common way malware reaches phones. Requirement: Apps are installed only from the official store (no alternative marketplaces, web distribution or unknown sources). How to fix on Android: (1) Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play. How to verify: Re-run the phone check; apps should come only from the official store (or the employer). Related controls: ISO 27001:2022 A.8.1 (User endpoint devices); ISO 27001:2022 A.8.7 (Protection against malware); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] Google Play Protect on Android: Google Play Protect on Android. Why it matters: Anti-malware software detects and blocks known malicious files and behaviour. Risk if missing: High - Without working anti-malware, ransomware and info-stealers can run unchallenged. Requirement: An enabled, up-to-date antivirus product is present. Android: Google Play Protect scanning is on. How to fix on Android: (1) Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'. How to verify: Re-run the audit; Antivirus should be Compliant. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] App updates on Android: App updates on Android. Why it matters: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Risk if missing: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages. Requirement: Installed third-party applications with a newer version available <= MAX_OUTDATED_APPS (Windows: winget; macOS: app versions; Linux: snap/flatpak and packages). Phones: automatic app updates are on. How to fix on Android: (1) Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all. How to verify: Re-run the audit; outdated applications should be 0 or fewer. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); ISO 27001:2022 A.8.19 (Installation of software on operational systems); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nQUESTION: How can I set up install unknown apps blocked on my Android?"}, {"role": "assistant", "content": "Official app stores review apps for malware; apps from other sources skip those checks. To fix it on Android [2]:\n1. Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\n\nTo check it worked: Re-run the phone check; apps should come only from the official store (or the employer)."}], "meta": {"task": "chat", "os": "android", "status": "fix", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: disabled\nREQUIRED: enabled\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"The device reports disabled, but the baseline requires enabled. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Password must meet complexity requirements = Enabled.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ConfigDataInstall=0, CriticalUpdateInstall=1\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware security updates is non-compliant\", \"finding\": \"Found: ConfigDataInstall=0, CriticalUpdateInstall=1. Required: automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall outbound policy is non-compliant\", \"finding\": \"Firewall outbound policy is no firewall instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: screen saver does not require sign-in for: FABRIKAM\\sali\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"Found: screen saver does not require sign-in for: FABRIKAM\\\\sali. Required: sign-in required to unlock. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: restic-backup.timer.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup is pending\", \"finding\": \"Found: date not readable. Confirm the last backup date in: restic-backup.timer.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 296 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 19 applications inventoried, 15 recognised (31 common applications covered). Google Chrome 152.0.7977.82 (259 CVEs, critical, actively exploited; fixed in 155.0.8059.39; e.g. CVE-2026-102299, CVE-2026-102301); Mozilla Firefox (x64 en-US) 156.0.0 (32 CVEs, critical; fixed in 157.0.0; e.g. CVE-2026-100756, CVE-2026-100767); 7-Zip 25.01 (x64) 25.01 (5 CVEs, high; fixed in 26.02; e.g. CVE-2026-14266, CVE-2026-48092). 316 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"Found: 296 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000151, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall outbound policy [firewall.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no firewall\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (linux):\n- ufw default allow outgoing (or the equivalent nftables output policy).\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall outbound policy is non-compliant\", \"finding\": \"The device reports no firewall, but the baseline requires allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"ufw default allow outgoing (or the equivalent nftables output policy).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access matches the required setting\", \"finding\": \"The device reports allow, which meets the requirement (allow).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates is non-compliant\", \"finding\": \"Malware signature updates is freshclam not running instead of automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 26.0.1 (macos)\nHOSTNAME: DEV-MacBook-58\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=37, Non-Compliant=9, NotApplicable=4, Pending=6, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Compliant | letters and numbers required\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5206\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5399, 15 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Compliant | on\n- firewall.inbound | Firewall inbound policy | Compliant | block (firewall on, stealth mode on)\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 180 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after 60 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (jdoe)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 241 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | check, download and install automatic\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 59 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 26.0.1\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | bzserv\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | screensharing\n- gdpr.telemetry | Diagnostic data minimised | Compliant | analytics sharing off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Jan\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Jan\n- aiact.risk_classification | AI system risk classification | Pending | Jan\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | restricted by profile\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- screen_lock.password: Medium\n- user.account_type: High\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- org.ai_register: Medium\n- org.ai_literacy: Medium\nSCORE (computed): compliant=37, assessed=46, percent=80\nTOP GAPS (computed, highest risk first):\n- user.account_type | Signed-in user account type | High | Administrator (jdoe)\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 241 critical/high\n- gdpr.updates_recent | Security updates installed recently | High | 59 days ago\n- gdpr.remote_access | Remote access services restricted | High | screensharing\n- screen_lock.password | Screen lock password requirement | Medium | password after 60 seconds\nFIX GUIDANCE (highest-risk gaps):\n- user.account_type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.updates_recent: Install pending updates (System Settings > General > Software Update, or softwareupdate -ia)."}, {"role": "assistant", "content": "{\"device\": \"DEV-MacBook-58 - macOS 26.0.1\", \"score\": {\"compliant\": 37, \"assessed\": 46, \"percent\": 80}, \"counts\": {\"Compliant\": 37, \"Non-Compliant\": 9, \"NotApplicable\": 4, \"Pending\": 6, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (jdoe)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"241 critical/high\"}, {\"check_id\": \"gdpr.updates_recent\", \"name\": \"Security updates installed recently\", \"risk\": \"High\", \"found\": \"59 days ago\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"screensharing\"}, {\"check_id\": \"screen_lock.password\", \"name\": \"Screen lock password requirement\", \"risk\": \"Medium\", \"found\": \"password after 60 seconds\"}], \"summary\": \"DEV-MacBook-58 meets 37 of 46 assessed checks (80%). The most serious gaps are Signed-in user account type, Known vulnerabilities in installed software and Security updates installed recently. 6 items are pending.\", \"next_steps\": [\"Signed-in user account type: Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Security updates installed recently: Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\", \"Answer the 3 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000567, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: USB debugging off [mobile.usb_debugging] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: on\nREQUIRED: off\nDETAIL: A connected computer can read and change data on the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: USB debugging lets a connected computer install apps and read data without the usual checks.\nrisk: Medium - Anyone who can plug the unlocked phone into a computer can copy data or install software.\nfix (android):\n- Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\nverify: Re-run the phone check; USB debugging should be off."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.usb_debugging\", \"status\": \"Non-Compliant\", \"title\": \"USB debugging off: does not meet the baseline\", \"finding\": \"USB debugging off is on instead of off. USB debugging lets a connected computer install apps and read data without the usual checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Anyone who can plug the unlocked phone into a computer can copy data or install software.\"}, \"recommendation\": [\"Settings › System › Developer options: turn off USB debugging, or turn off Developer options completely.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; USB debugging should be off.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.usb_debugging", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Tamper protection [malware.tamper] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: Defender in Passive Mode\nREQUIRED: on\nDETAIL: Another antivirus is primary: Sophos Endpoint\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Tamper protection stops malware or users switching off Defender.\nrisk: Medium - Attackers commonly disable antivirus before deploying ransomware.\nfix (windows):\n- Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell).\nverify: Re-run the audit; Tamper protection should be on.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.tamper\", \"status\": \"NotApplicable\", \"title\": \"Tamper protection does not apply here\", \"finding\": \"Not applicable on this device: Another antivirus is primary: Sophos Endpoint.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.tamper", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: usage & diagnostics off\nREQUIRED: usage & diagnostics off\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Compliant\", \"title\": \"Usage & diagnostics sharing minimised is compliant\", \"finding\": \"Usage & diagnostics sharing minimised is set to usage & diagnostics off, in line with the required usage & diagnostics off. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iOS 26.7 (ios)\nAI TOOLS FOUND: ChatGPT, Gemini, Grammarly\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Gemini: Cloud AI assistant; provider Google (US); data: Provider cloud; default risk minimal; GDPR: Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\n- Grammarly: AI writing assistant (keyboard and browser); provider Grammarly (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Reads everything typed where it is enabled, including personal data; use the business plan with a data processing agreement or disable it in work apps.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Gemini\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Workspace version under the organisation's Google data processing terms; consumer use is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Grammarly\", \"kind\": \"AI writing assistant (keyboard and browser)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Reads everything typed where it is enabled, including personal data; use the business plan with a data processing agreement or disable it in work apps.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Gemini, Grammarly.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Gemini\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Current state: Gemini. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000266, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain firewall [firewall.domain.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Domain firewall is non-compliant\", \"finding\": \"Domain firewall is off instead of on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apps only from the App Store\", \"finding\": \"The user reports apps from outside the App Store, but the baseline requires App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001257, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nHOSTNAME: dev-lt-95\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=29, Non-Compliant=16, NotApplicable=5, Pending=5, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 180 days\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Non-Compliant | minclass=2, required classes=1\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 30 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | Bitdefender Endpoint Security\n- malware.service | ClamAV malware protection service | NotApplicable | not used\n- malware.signature_updates | Malware signature updates | NotApplicable | not used\n- malware.signatures | Malware signatures | NotApplicable | not used\n- malware.on_access | On-access malware protection | Compliant | Bitdefender Endpoint Security\n- firewall.enabled | Firewall | Compliant | firewalld active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Non-Compliant | off for: lmuller, asmith\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: lmuller, asmith\n- usb.device_access | USB storage device access | Non-Compliant | deny\n- usb.read | USB storage read | Non-Compliant | deny\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | sali, mkhan\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (lmuller)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 45 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 89 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 56 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Red Hat Enterprise Linux 9.6 (Plow)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Error | not read\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Hugging Face models\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Hugging Face models\n- aiact.risk_classification | AI system risk classification | Pending | Hugging Face models\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.complexity: Medium\n- screen_lock.enabled: Medium\n- screen_lock.password: Medium\n- usb.device_access: Low\n- usb.read: Low\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- aiact.unapproved: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=29, assessed=45, percent=64\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee administrator accounts | High | sali, mkhan\n- user.account_type | Signed-in user account type | High | Administrator (lmuller)\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 89 critical/high\n- gdpr.updates_recent | Security updates installed recently | High | 56 days ago\n- gdpr.backup_configured | Backup configured | High | none\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- user.account_type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"dev-lt-95 - Red Hat Enterprise Linux 9.6 (Plow) 9.6\", \"score\": {\"compliant\": 29, \"assessed\": 45, \"percent\": 64}, \"counts\": {\"Compliant\": 29, \"Non-Compliant\": 16, \"NotApplicable\": 5, \"Pending\": 5, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"sali, mkhan\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (lmuller)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"89 critical/high\"}, {\"check_id\": \"gdpr.updates_recent\", \"name\": \"Security updates installed recently\", \"risk\": \"High\", \"found\": \"56 days ago\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}], \"summary\": \"dev-lt-95 meets 29 of 45 assessed checks (64%). The most serious gaps are Unauthorized employee administrator accounts, Signed-in user account type and Known vulnerabilities in installed software. 1 checks could not be read. 5 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Signed-in user account type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 3 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: swipe or none\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: There is no secret needed to unlock the phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock strength\", \"finding\": \"Screen lock strength is swipe or none instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.7 checked against NVD iOS records. iOS 26.7 (15 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-43688). 32 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Known vulnerabilities in iOS\", \"finding\": \"Found: 15 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001145, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"Found: 5 minutes. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutwindow:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000818, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"The device reports USB storage writable without enforced encryption, but the baseline requires no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password age [password.min_age] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MIN_DAYS 0\nREQUIRED: >= 1 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A minimum age stops users cycling through passwords immediately to get back to an old one.\nrisk: Low - Without it, the password history rule can be bypassed.\nfix (linux):\n- Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\nverify: Re-run the audit; Minimum password age should read 1 day(s) or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_age\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password age: does not meet the baseline\", \"finding\": \"The device reports PASS_MIN_DAYS 0, but the baseline requires >= 1 days. A minimum age stops users cycling through passwords immediately to get back to an old one.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Without it, the password history rule can be bypassed.\"}, \"recommendation\": [\"Set PASS_MIN_DAYS 1 in /etc/login.defs and apply to existing accounts: chage --mindays 1 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password age should read 1 day(s) or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.min_age", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Date & time set automatically [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: set by hand\nREQUIRED: set automatically\nDETAIL: Wrong time breaks certificate checks and makes logs unreliable.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (ios):\n- Settings › General › Date & Time: turn on Set Automatically.\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Date & time set automatically\", \"finding\": \"The user reports set by hand, but the baseline requires set automatically. Accurate clocks are needed to put events in the right order during an investigation. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"Settings › General › Date & Time: turn on Set Automatically.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"AI transparency to users is no instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen lock strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit PIN\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (android):\n- Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock strength\", \"finding\": \"Screen lock strength is 4-digit PIN instead of >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Security & privacy › Device unlock › Screen lock: choose a PIN of at least 6 digits or a password instead of a pattern.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (sali)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"The device reports Administrator (sali), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: clamav-daemon\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Compliant\", \"title\": \"ClamAV malware protection service matches the required setting\", \"finding\": \"Found: clamav-daemon. This satisfies the baseline requirement of running.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Cursor (AI tool): Cursor: AI code editor, provider Anysphere (US), data location: Code sent to the provider and its model providers unless privacy mode is enforced. Default EU AI Act risk class for general office use: minimal. GDPR: Enforce privacy mode through a business plan and sign a data processing agreement. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[2] Claude Code (AI tool): Claude Code: AI coding assistant (command line), provider Anthropic (US), data location: Code and prompts sent to the provider's cloud. Default EU AI Act risk class for general office use: minimal. GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n[3] Privacy: where do audit results go?: Nowhere. Evidence and reports stay on the device. The app downloads only two things: the AI auditor model and a signed list of known vulnerabilities. Nothing about the device is uploaded. Chat history is stored only in this browser.\n[4] Windsurf (AI tool): Windsurf: AI code editor, provider US-based provider, data location: Code sent to the provider's cloud. Default EU AI Act risk class for general office use: minimal. GDPR: Use a business plan with a data processing agreement and zero-retention settings. Whether it is allowed at work depends on the organisation's approved AI tools list and on the purpose: some purposes make any AI tool high-risk or prohibited.\n\nMY AUDIT: none\n\nMY DEVICE: Windows 11 (windows) on a Windows PC, Firefox 143; detected by the browser, security settings not checked.\n\nQUESTION: Is my data sent anywhere?"}, {"role": "assistant", "content": "Nowhere. Evidence and reports stay on the device. The app downloads only two things: the AI auditor model and a signed list of known vulnerabilities. Nothing about the device is uploaded. Chat history is stored only in this browser. [3]"}], "meta": {"task": "chat", "os": "any", "status": "faq", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: download and install on\nDETAIL: Security updates are not installed automatically.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic iOS updates is non-compliant\", \"finding\": \"Automatic iOS updates is off instead of download and install on. Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000501, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked is non-compliant\", \"finding\": \"The user reports Always, but the baseline requires When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Enforce password history = 24.\n- For a stand-alone device: net accounts /uniquepw:24\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history matches the required setting\", \"finding\": \"Found: 24. This satisfies the baseline requirement of >= 24.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: GPT4All\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: GPT4All. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My iPhone [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (ios):\n- Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My iPhone is non-compliant\", \"finding\": \"Find My iPhone is off instead of on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › [your name] › Find My › Find My iPhone: turn on, together with Send Last Location.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware protection signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect 5364, 7 days old\nREQUIRED: <= 30 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (macos):\n- Keep 'Install Security Responses and system files' on (Software Update > Automatic Updates); run softwareupdate --background to fetch now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Compliant\", \"title\": \"Malware protection signatures is compliant\", \"finding\": \"The device reports XProtect 5364, 7 days old, which meets the requirement (<= 30 days old).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apps only from the App Store\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4 outdated\nREQUIRED: <= 0 outdated\nDETAIL: VLC media player VideoLAN.VLC 31.8.535 31.4.32 winget; Notepad++ (64-bit x64) Notepad++.Notepad++ 97.1.264 98.7.41 winget; Google Chrome Google.Chrome 43.6.678 43.6.942 winget; Git Git.Git 127.6.913 127.6.0 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"Outdated third-party applications is 4 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000514, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Perplexity, ChatGPT, Claude\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: needs a follow-up step\", \"finding\": \"Current state: Perplexity, ChatGPT, Claude. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: running\nDETAIL: Value could not be read.\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Error\", \"title\": \"ClamAV malware protection service could not be read\", \"finding\": \"The collector could not read this setting (Value could not be read.), so compliance is unknown.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 12 (android)\nAI TOOLS FOUND: ChatGPT\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification: not applicable\", \"finding\": \"AI system risk classification is not assessed here: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Answer ORG_AI_LITERACY in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: rdp\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (windows):\n- Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"The device reports rdp, but the baseline requires only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Turn off the listed services if not approved: disable Remote Desktop (fDenyTSConnections = 1), stop and disable sshd/WinRM, remove unneeded shares, and uninstall unapproved tools such as TeamViewer, AnyDesk or Chrome Remote Desktop.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: 4 days ago\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Compliant\", \"title\": \"Recent backup: meets the baseline\", \"finding\": \"Found: 4 days ago. This satisfies the baseline requirement of <= 7 days ago.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LAPTOP-3TD06P0\\rschmit\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee local administrator accounts\", \"finding\": \"Unauthorized employee local administrator accounts is LAPTOP-3TD06P0\\\\rschmit instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: iOS 26.7 (ios)\nHOSTNAME: LEGAL-iPhone-57\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=9, Non-Compliant=10, NotApplicable=0, Pending=10, Error=0\nEVIDENCE: 2 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- screen_lock.enabled | Passcode | Compliant | passcode set\n- screen_lock.timeout | Auto-Lock | Compliant | Auto-Lock 5 minutes\n- mobile.passcode_strength | Passcode strength | Compliant | alphanumeric\n- mobile.jailbreak_root | Not jailbroken | Compliant | not jailbroken (user confirmed)\n- mobile.app_sources | Apps only from the App Store | Non-Compliant | apps from outside the App Store\n- mobile.find_my_device | Find My iPhone | Non-Compliant | off\n- encryption.enabled | Data protection (encryption) | Compliant | on (passcode set)\n- vuln.pending_updates | iOS security updates | Non-Compliant | iOS 26.7 (update to 26.7.1 available)\n- vuln.outdated_apps | App updates | Non-Compliant | automatic app updates off\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | Non-Compliant | 15 critical/high\n- gdpr.auto_updates | Automatic iOS updates | Compliant | download and install on\n- gdpr.os_supported | Supported operating system version | Compliant | iOS 26.7\n- gdpr.time_sync | Date & time set automatically | Non-Compliant | set by hand\n- gdpr.backup_configured | iCloud Backup | Non-Compliant | no backup\n- gdpr.telemetry | Analytics sharing minimised | Non-Compliant | analytics sharing on\n- gdpr.lock_screen_previews | Notification previews hidden when locked | Compliant | Never\n- aiact.inventory | AI software inventory | Compliant | ChatGPT, Gemini, Grammarly\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT, Gemini, Grammarly\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT, Gemini, Grammarly\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Non-Compliant | ChatGPT extension on\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Pending | not answered\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- mobile.app_sources: High\n- mobile.find_my_device: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.time_sync: Low\n- gdpr.backup_configured: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- aiact.os_ai_features: Medium\nSCORE (computed): compliant=9, assessed=19, percent=47\nTOP GAPS (computed, highest risk first):\n- mobile.app_sources | Apps only from the App Store | High | apps from outside the App Store\n- vuln.pending_updates | iOS security updates | High | iOS 26.7 (update to 26.7.1 available)\n- vuln.known_vulnerabilities | Known vulnerabilities in iOS | High | 15 critical/high\n- gdpr.backup_configured | iCloud Backup | High | no backup\n- mobile.find_my_device | Find My iPhone | Medium | off\nFIX GUIDANCE (highest-risk gaps):\n- mobile.app_sources: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\n- vuln.pending_updates: Settings › General › Software Update: install the latest iOS version offered.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use."}, {"role": "assistant", "content": "{\"device\": \"LEGAL-iPhone-57 - iOS 26.7\", \"score\": {\"compliant\": 9, \"assessed\": 19, \"percent\": 47}, \"counts\": {\"Compliant\": 9, \"Non-Compliant\": 10, \"NotApplicable\": 0, \"Pending\": 10, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"mobile.app_sources\", \"name\": \"Apps only from the App Store\", \"risk\": \"High\", \"found\": \"apps from outside the App Store\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"iOS security updates\", \"risk\": \"High\", \"found\": \"iOS 26.7 (update to 26.7.1 available)\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in iOS\", \"risk\": \"High\", \"found\": \"15 critical/high\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"iCloud Backup\", \"risk\": \"High\", \"found\": \"no backup\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My iPhone\", \"risk\": \"Medium\", \"found\": \"off\"}], \"summary\": \"LEGAL-iPhone-57 meets 9 of 19 assessed checks (47%). The most serious gaps are Apps only from the App Store, iOS security updates and Known vulnerabilities in iOS. 10 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Apps only from the App Store: Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\", \"iOS security updates: Settings › General › Software Update: install the latest iOS version offered.\", \"Known vulnerabilities in iOS: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Answer the 8 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "ios", "audit": 7000079, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after off seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"Screen lock password requirement is password after off seconds instead of password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000765, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nHOSTNAME: DESKTOP-H6TXNO4\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=30, Non-Compliant=29, NotApplicable=3, Pending=2, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Non-Compliant | 0\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Compliant | 5\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | Windows Defender (on, up to date)\n- malware.service | Microsoft Defender Antimalware | Compliant | service True, antivirus True\n- malware.realtime | Real-time protection | Compliant | on\n- malware.on_access | On-access protection | Compliant | on\n- malware.tamper | Tamper protection | Non-Compliant | off\n- malware.signatures | Security intelligence | Compliant | 0 days old (version 1.441.521.0)\n- firewall.domain.enabled | Domain firewall | Compliant | on\n- firewall.domain.inbound | Domain inbound | Compliant | block\n- firewall.domain.outbound | Domain outbound | Non-Compliant | block\n- firewall.private.enabled | Private firewall | Compliant | on\n- firewall.private.inbound | Private inbound | Compliant | block\n- firewall.private.outbound | Private outbound | Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | password-protected screen saver on for all signed-in users\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | 1200 seconds (machine inactivity limit)\n- screen_lock.password | Screen lock password requirement | Compliant | screen saver requires sign-in for all signed-in users\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an enabled administrator\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-H6TXNO4\\akaya\n- user.account_type | Signed-in user account type | Compliant | Standard (DESKTOP-H6TXNO4\\tnguyen)\n- encryption.feature | BitLocker | Non-Compliant | off\n- encryption.enabled | Encryption | Non-Compliant | FullyDecrypted (0%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, None\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 4 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 144 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | AUOptions=4\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 57 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Pro build 26200\n- gdpr.security_logging | Security event logging | Non-Compliant | logon: none; account management: none\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | BitLocker To Go required before writing\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | AllowTelemetry=3 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Compliant | Recall disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Non-Compliant | partial\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | partial\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.expiration: Low\n- password.min_age: Low\n- password.complexity: Medium\n- malware.tamper: Medium\n- firewall.domain.outbound: Low\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- encryption.feature: High\n- encryption.enabled: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.telemetry: Low\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=30, assessed=59, percent=51\nTOP GAPS (computed, highest risk first):\n- firewall.public.enabled | Public firewall | High | off\n- firewall.public.inbound | Public inbound | High | allow\n- admin.unauthorized | Unauthorized employee local administrator accounts | High | DESKTOP-H6TXNO4\\akaya\n- encryption.feature | BitLocker | High | off\n- encryption.enabled | Encryption | High | FullyDecrypted (0%)\nFIX GUIDANCE (highest-risk gaps):\n- firewall.public.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.public.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- admin.unauthorized: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool)."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-H6TXNO4 - Microsoft Windows 11 Pro 10.0.26200\", \"score\": {\"compliant\": 30, \"assessed\": 59, \"percent\": 51}, \"counts\": {\"Compliant\": 30, \"Non-Compliant\": 29, \"NotApplicable\": 3, \"Pending\": 2, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.public.enabled\", \"name\": \"Public firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.public.inbound\", \"name\": \"Public inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee local administrator accounts\", \"risk\": \"High\", \"found\": \"DESKTOP-H6TXNO4\\\\akaya\"}, {\"check_id\": \"encryption.feature\", \"name\": \"BitLocker\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"encryption.enabled\", \"name\": \"Encryption\", \"risk\": \"High\", \"found\": \"FullyDecrypted (0%)\"}], \"summary\": \"DESKTOP-H6TXNO4 meets 30 of 59 assessed checks (51%). The most serious gaps are Public firewall, Public inbound and Unauthorized employee local administrator accounts. 2 items are pending.\", \"next_steps\": [\"Public firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Public inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Unauthorized employee local administrator accounts: Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\", \"Answer the 2 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (windows):\n- Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup is non-compliant\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Check the backup tool's history and run a backup now; fix whatever stopped it (full destination, signed-out OneDrive, stopped agent).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 37 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-08-26 15:06:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"The device reports 37 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (macos):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Breach response procedure\", \"finding\": \"The device reports no, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: On-access malware protection\", \"finding\": \"On-access malware protection is not running instead of on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"Found: partial. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall [firewall.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: firewalld active\nREQUIRED: active\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (linux):\n- Debian/Ubuntu: ufw default deny incoming; ufw default allow outgoing; ufw enable. RHEL: systemctl enable --now firewalld.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.enabled\", \"status\": \"Compliant\", \"title\": \"Firewall: meets the baseline\", \"finding\": \"The device reports firewalld active, which meets the requirement (active).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: guest login 1, guest file sharing 0\nREQUIRED: guest login off\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (macos):\n- sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Guest account disabled\", \"finding\": \"Guest account disabled is guest login 1, guest file sharing 0 instead of guest login off. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"sudo defaults write /Library/Preferences/com.apple.loginwindow GuestEnabled -bool false, and turn off guest access to shared folders.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nDETAIL: AI apps listed by the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (ios):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"The user reports no AI tools found, which meets the requirement (inventory taken). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000967, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Compliant\", \"title\": \"Record of processing activities is compliant\", \"finding\": \"Record of processing activities is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000067, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_TRANSPARENCY in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_TRANSPARENCY in organisational.conf.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: ChatGPT. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 13.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001175, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Find My Device [mobile.find_my_device] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nDETAIL: A lost phone cannot be located, locked or erased remotely.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Locating, locking and erasing a lost phone limits a personal data breach.\nrisk: Medium - A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\nfix (android):\n- Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\nverify: Re-run the phone check; Find My should be on."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.find_my_device\", \"status\": \"Non-Compliant\", \"title\": \"Find My Device is non-compliant\", \"finding\": \"The user reports off, but the baseline requires on. Locating, locking and erasing a lost phone limits a personal data breach. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost phone cannot be wiped, so its data may have to be reported as a breach (GDPR Art. 33).\"}, \"recommendation\": [\"Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the phone check; Find My should be on.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.find_my_device", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Ollama, Perplexity, Jan\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (macos):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Found: Ollama, Perplexity, Jan. Classified by the auditor model (Caveat classifier) in a later step.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 90 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-04 Cumulative Update for Windows 11 (2026-07-06)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently is non-compliant\", \"finding\": \"Security updates installed recently is 90 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-J0Q0Y2I\\tnguyen\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Unauthorized employee local administrator accounts\", \"finding\": \"Found: DESKTOP-J0Q0Y2I\\\\tnguyen. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: BitLocker recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: key protectors: Tpm, RecoveryPassword\nREQUIRED: recovery password present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (windows):\n- Add a recovery password: Add-BitLockerKeyProtector -MountPoint C: -RecoveryPasswordProtector, and back it up to Entra ID (BackupToAAD-BitLockerKeyProtector) or Active Directory.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Compliant\", \"title\": \"BitLocker recovery matches the required setting\", \"finding\": \"Found: key protectors: Tpm, RecoveryPassword. This satisfies the baseline requirement of recovery password present.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 3 outdated\nREQUIRED: <= 0 outdated\nDETAIL: 7-Zip 24.08 (x64) 7zip.7zip 79.3.91 80.9.74 winget; Adobe Acrobat Reader (64-bit) Adobe.Acrobat.Reader.64-bit 116.1.923 116.5.744 winget; Notepad++ (64-bit x64) Notepad++.Notepad++ 99.2.389 99.8.300 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"Outdated third-party applications is 3 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Compliant\", \"title\": \"Domain outbound matches the required setting\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude Code\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: Claude Code. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved tools with the package manager and restrict installation rights.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register is non-compliant\", \"finding\": \"AI register is no instead of yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Google Play Protect: does not meet the baseline\", \"finding\": \"Google Play Protect is Play Protect scanning off instead of on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nHOSTNAME: DESKTOP-J3GFH4F\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=16, Non-Compliant=37, NotApplicable=8, Pending=3, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Non-Compliant | 5\n- password.expiration | Password expiration | Non-Compliant | never\n- password.min_age | Minimum password age | Non-Compliant | 0 days\n- password.complexity | Password complexity | Non-Compliant | disabled\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout configured\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | Bitdefender Endpoint Security (on, up to date); Windows Defender (off, up to date)\n- malware.service | Microsoft Defender Antimalware | NotApplicable | Defender in Passive Mode\n- malware.realtime | Real-time protection | NotApplicable | Defender in Passive Mode\n- malware.on_access | On-access protection | NotApplicable | Defender in Passive Mode\n- malware.tamper | Tamper protection | NotApplicable | Defender in Passive Mode\n- malware.signatures | Security intelligence | NotApplicable | Defender in Passive Mode\n- firewall.domain.enabled | Domain firewall | Non-Compliant | off\n- firewall.domain.inbound | Domain inbound | Non-Compliant | allow\n- firewall.domain.outbound | Domain outbound | Non-Compliant | block\n- firewall.private.enabled | Private firewall | Non-Compliant | off\n- firewall.private.inbound | Private inbound | Non-Compliant | allow\n- firewall.private.outbound | Private outbound | Non-Compliant | allow\n- firewall.public.enabled | Public firewall | Non-Compliant | off\n- firewall.public.inbound | Public inbound | Non-Compliant | allow\n- firewall.public.outbound | Public outbound | Non-Compliant | block\n- screen_lock.enabled | Screen lock | Non-Compliant | no machine inactivity limit; no signed-in user to check\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | no machine inactivity limit; screen saver off\n- screen_lock.password | Screen lock password requirement | Compliant | machine inactivity limit locks the session\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | CONTOSO\\itadmin\n- admin.unauthorized | Unauthorized employee local administrator accounts | Non-Compliant | DESKTOP-J3GFH4F\\mkhan, DESKTOP-J3GFH4F\\Administrator\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (CONTOSO\\jdoe)\n- encryption.feature | BitLocker | Compliant | on\n- encryption.enabled | Encryption | Compliant | FullyEncrypted (100%)\n- encryption.protection | BitLocker protection | Non-Compliant | protection Off, XtsAes128\n- encryption.recovery | BitLocker recovery | Non-Compliant | key protectors: Tpm\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 4 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 222 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | Windows Update service disabled\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 226 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Microsoft Windows 11 Enterprise build 26100\n- gdpr.security_logging | Security event logging | Compliant | logon: success+failure; account management: success+failure\n- gdpr.time_sync | Time synchronisation | Compliant | type NT5DS, service Running/Manual\n- gdpr.backup_configured | Backup configured | Non-Compliant | none found\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB drives writable without encryption\n- gdpr.guest_account | Guest account disabled | Compliant | disabled\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | AllowTelemetry=3 (Settings)\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Windows Recall / Copilot data analysis disabled | Non-Compliant | Recall feature Disabled, not disabled by policy\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Pending | not answered\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Non-Compliant | partial\nRISK LEVELS (from guidance):\n- password.history: Low\n- password.expiration: Low\n- password.min_age: Low\n- password.complexity: Medium\n- lockout.threshold: Medium\n- lockout.duration: Low\n- firewall.domain.enabled: High\n- firewall.domain.inbound: High\n- firewall.domain.outbound: Low\n- firewall.private.enabled: High\n- firewall.private.inbound: High\n- firewall.private.outbound: Low\n- firewall.public.enabled: High\n- firewall.public.inbound: High\n- firewall.public.outbound: Low\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- usb.write: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.protection: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.updates_recent: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.telemetry: Low\n- aiact.os_ai_features: Medium\n- org.dpia: Medium\n- org.processor_agreements: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=16, assessed=53, percent=30\nTOP GAPS (computed, highest risk first):\n- firewall.domain.enabled | Domain firewall | High | off\n- firewall.domain.inbound | Domain inbound | High | allow\n- firewall.private.enabled | Private firewall | High | off\n- firewall.private.inbound | Private inbound | High | allow\n- firewall.public.enabled | Public firewall | High | off\nFIX GUIDANCE (highest-risk gaps):\n- firewall.domain.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\n- firewall.domain.inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\n- firewall.private.enabled: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune."}, {"role": "assistant", "content": "{\"device\": \"DESKTOP-J3GFH4F - Microsoft Windows 11 Enterprise 10.0.26100\", \"score\": {\"compliant\": 16, \"assessed\": 53, \"percent\": 30}, \"counts\": {\"Compliant\": 16, \"Non-Compliant\": 37, \"NotApplicable\": 8, \"Pending\": 3, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.domain.enabled\", \"name\": \"Domain firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.domain.inbound\", \"name\": \"Domain inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.private.enabled\", \"name\": \"Private firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.private.inbound\", \"name\": \"Private inbound\", \"risk\": \"High\", \"found\": \"allow\"}, {\"check_id\": \"firewall.public.enabled\", \"name\": \"Public firewall\", \"risk\": \"High\", \"found\": \"off\"}], \"summary\": \"DESKTOP-J3GFH4F meets 16 of 53 assessed checks (30%). The most serious gaps are Domain firewall, Domain inbound and Private firewall. 3 items are pending.\", \"next_steps\": [\"Domain firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Domain inbound: Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\", \"Private firewall: Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\", \"Answer the 3 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Where to find Android security update on Android: Android security update on Android: What is the Android security update date? Where to look: Settings › About phone › Android version › Android security update (or Settings › Security & privacy › Updates)\n[2] Where to find App updates on Android: App updates on Android: Are apps updated automatically? Where to look: Play Store › your profile picture › Settings › Network preferences › Auto-update apps\n[3] Automatic system updates on Android: Automatic system updates on Android. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on Android: (1) Settings › System › Software updates: turn on automatic download (names vary by phone maker). How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Where to find Automatic system updates on Android: Automatic system updates on Android: Are system updates downloaded automatically? Where to look: Settings › System › Software updates (names vary: 'Auto download over Wi-Fi', 'Automatic system updates')\n\nMY AUDIT: none\n\nMY DEVICE: Android 14 (android) on an Android phone, Chrome 141; detected by the browser, security settings not checked.\n\nQUESTION: Where do I find automatic system updates on my Android?"}, {"role": "assistant", "content": "Go to Settings › System › Software updates (names vary: 'Auto download over Wi-Fi', 'Automatic system updates') [4]"}], "meta": {"task": "chat", "os": "android", "status": "phone", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Removable media encryption is USB drives writable without encryption instead of no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Private outbound\", \"finding\": \"Private outbound is allow instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"Found: not set. Required: >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Account lockout threshold [lockout.threshold] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: 1-5\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Locking an account after a few wrong passwords stops online guessing.\nrisk: Medium - Unlimited attempts allow brute-force attacks against local accounts.\nfix (macos):\n- Deploy a Passcode configuration profile with maxFailedAttempts = 5.\nverify: Re-run the audit; Account lockout threshold should be between 1 and 5."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.threshold\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout threshold is non-compliant\", \"finding\": \"Found: not set. Required: 1-5. Locking an account after a few wrong passwords stops online guessing.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Unlimited attempts allow brute-force attacks against local accounts.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxFailedAttempts = 5.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout threshold should be between 1 and 5.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.threshold", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no FileVault\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Disk encryption recovery\", \"finding\": \"Disk encryption recovery is no FileVault instead of recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000729, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: root filesystem not encrypted\nREQUIRED: system disk encrypted\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (linux):\n- Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption is non-compliant\", \"finding\": \"Encryption is root filesystem not encrypted instead of system disk encrypted. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Disk encryption (LUKS) is usually chosen at installation; plan a reinstall with encryption, or an offline re-encryption with cryptsetup reencrypt after a full backup.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Grammarly, Microsoft Copilot\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Unapproved AI tools (shadow AI) is ChatGPT, Grammarly, Microsoft Copilot instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: external drives writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (macos):\n- Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Found: external drives writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Restrict external storage writes with your MDM or an endpoint DLP product and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Public inbound: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: decision pending\", \"finding\": \"Found: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Screen timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Screen timeout 10 minutes\nREQUIRED: <= 300 seconds\nDETAIL: The phone stays unlocked for 10 minutes when not in use.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (android):\n- Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen timeout\", \"finding\": \"The user reports Screen timeout 10 minutes, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display › Screen timeout: choose 5 minutes or less, and set 'Lock after screen timeout' to Immediately where the phone offers it.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Error\nFOUND: not read\nREQUIRED: deny\nDETAIL: Requires root: run the collector with sudo.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny.\nstatus note: The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Error\", \"title\": \"Firewall inbound policy could not be read\", \"finding\": \"This check returned an error: Requires root: run the collector with sudo. Compliance cannot be confirmed.\", \"risk\": {\"level\": \"Unknown\", \"reason\": \"The setting could not be read.\"}, \"recommendation\": [\"The value could not be read, so compliance is unknown. Re-run the collector as administrator/root (Windows: elevated PowerShell or SYSTEM; Linux/macOS: sudo) to collect the evidence.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000167, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Error"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Meta AI\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Found: ChatGPT, Meta AI. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (ios):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Found: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: minclass=0, required classes=0\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password complexity\", \"finding\": \"The device reports minclass=0, required classes=0, but the baseline requires 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 160 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iPadOS 18.7.9 checked against NVD iOS records. iPadOS 18.7.9 (160 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2025-31255, CVE-2025-43323). 285 medium/low.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS: does not meet the baseline\", \"finding\": \"The user reports 160 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: disabled\nREQUIRED: disabled\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (windows):\n- Disable the built-in Guest account: Get-LocalUser | Where-Object { $_.SID -like '*-501' } | Disable-LocalUser\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled: meets the baseline\", \"finding\": \"The device reports disabled, which meets the requirement (disabled).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities is non-compliant\", \"finding\": \"Record of processing activities is partial instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"USB storage write is allow instead of deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nHOSTNAME: mkt-ubuntu-24\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=27, Non-Compliant=21, NotApplicable=7, Pending=1, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 14\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 99999\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=0, required classes=4\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Non-Compliant | no lockout\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | CrowdStrike Falcon\n- malware.service | ClamAV malware protection service | NotApplicable | not used\n- malware.signature_updates | Malware signature updates | NotApplicable | not used\n- malware.signatures | Malware signatures | NotApplicable | not used\n- malware.on_access | On-access malware protection | Compliant | CrowdStrike Falcon\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Non-Compliant | off for: rschmit\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: rschmit\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (rschmit)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 15 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 42 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 49 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Compliant | 19 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | rdp, chromeremotedesktop\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | no\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- screen_lock.enabled: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- admin.approved: Medium\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.ropa: High\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.human_oversight: Medium\nSCORE (computed): compliant=27, assessed=48, percent=56\nTOP GAPS (computed, highest risk first):\n- vuln.pending_updates | Pending security updates | High | 15 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 49 critical/high\n- gdpr.os_supported | Supported operating system version | High | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.backup_configured | Backup configured | High | none\n- gdpr.remote_access | Remote access services restricted | High | rdp, chromeremotedesktop\nFIX GUIDANCE (highest-risk gaps):\n- vuln.pending_updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.os_supported: Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance."}, {"role": "assistant", "content": "{\"device\": \"mkt-ubuntu-24 - Red Hat Enterprise Linux 8.10 (Ootpa) 8.10\", \"score\": {\"compliant\": 27, \"assessed\": 48, \"percent\": 56}, \"counts\": {\"Compliant\": 27, \"Non-Compliant\": 21, \"NotApplicable\": 7, \"Pending\": 1, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"15 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"49 critical/high\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Red Hat Enterprise Linux 8.10 (Ootpa)\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"rdp, chromeremotedesktop\"}], \"summary\": \"mkt-ubuntu-24 meets 27 of 48 assessed checks (56%). The most serious gaps are Pending security updates, Known vulnerabilities in installed software and Supported operating system version. 1 items are pending.\", \"next_steps\": [\"Pending security updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Supported operating system version: Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutwindow:15\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Lockout observation period\", \"finding\": \"Found: no lockout configured. Required: >= 15 minutes. Failed attempts must be counted over a long enough window to detect slow guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short window lets attackers pace their guesses to avoid lockout.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Reset account lockout counter after = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutwindow:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Lockout observation period should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_ROPA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answer ORG_ROPA in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Remote access services restricted [gdpr.remote_access] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: smb, rdp\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.6.7 Remote working: Protect information accessed, processed or stored outside the organisation's premises.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Remote access services let others reach the device over the network; only approved ones should be on.\nrisk: High - Unapproved remote access tools are a common entry point for attackers and support scams.\nfix (linux):\n- Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\nverify: Re-run the audit; only approved services (none) should remain.\nnote: If a listed service is needed, add it to APPROVED_REMOTE_SERVICES in baseline.conf and secure it (multi-factor authentication, restricted to the VPN)."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.remote_access\", \"status\": \"Non-Compliant\", \"title\": \"Remote access services restricted: does not meet the baseline\", \"finding\": \"Found: smb, rdp. Required: only: none. Remote access services let others reach the device over the network; only approved ones should be on.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unapproved remote access tools are a common entry point for attackers and support scams.\"}, \"recommendation\": [\"Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.6.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; only approved services (none) should remain.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.remote_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (linux):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Non-Compliant\", \"title\": \"Human oversight for high-risk AI is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk AI systems need trained people who can understand, monitor and override them.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (linux):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Record of processing activities\", \"finding\": \"Record of processing activities is partial instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Google Play Protect is non-compliant\", \"finding\": \"Google Play Protect is Play Protect scanning off instead of on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT, Claude\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (android):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Unapproved AI tools (shadow AI) is ChatGPT, Claude instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"Found: no. Required: yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Notification previews hidden when locked is non-compliant\", \"finding\": \"The user reports Always, but the baseline requires When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001100, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an enabled administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (windows):\n- Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Approved IT administrator\", \"finding\": \"The device reports none of the approved accounts is an enabled administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Add the approved IT administrator account (listed in APPROVED_ADMINS) to the local Administrators group, ideally managed with Windows LAPS or Intune.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Analytics sharing minimised: does not meet the baseline\", \"finding\": \"The user reports analytics sharing on, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001194, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not required for: lmuller\nREQUIRED: password required\nDETAIL: Users checked: lmuller\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (linux):\n- Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock password requirement\", \"finding\": \"Screen lock password requirement is not required for: lmuller instead of password required. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Keep org.gnome.desktop.screensaver lock-enabled true and org.gnome.desktop.lockdown disable-lock-screen false.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (android):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: FileVault off\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Encryption: does not meet the baseline\", \"finding\": \"The device reports FileVault off, but the baseline requires FileVault on. Full-disk encryption keeps data unreadable if the device is lost or stolen.\", \"risk\": {\"level\": \"High\", \"reason\": \"A lost unencrypted laptop is a personal data breach that may have to be reported.\"}, \"recommendation\": [\"Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"], \"verification\": \"Re-run the audit; Encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000397, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (ewagner)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"The device reports Administrator (ewagner), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after 60 seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement: does not meet the baseline\", \"finding\": \"The device reports password after 60 seconds, but the baseline requires password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements: does not meet the baseline\", \"finding\": \"Data processing agreements is partial instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Auto-Lock\", \"finding\": \"Auto-Lock is Auto-Lock Never instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Red Hat Enterprise Linux 8.10 (Ootpa)\nREQUIRED: RHEL 9 or later\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (linux):\n- Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Supported operating system version is non-compliant\", \"finding\": \"Supported operating system version is Red Hat Enterprise Linux 8.10 (Ootpa) instead of RHEL 9 or later. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Upgrade to a supported release (e.g. Ubuntu 22.04 or later) or enrol in extended security maintenance.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (macos):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Answer ORG_HUMAN_OVERSIGHT in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (linux):\n- Add pam_pwhistory to the PAM password stack before pam_unix (e.g. 'password requisite pam_pwhistory.so remember=24 use_authtok').\n- Or set remember = 24 in /etc/security/pwhistory.conf where supported.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history is compliant\", \"finding\": \"Password history is set to 24, in line with the required >= 24.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nHOSTNAME: legal-dev-07\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=32, Non-Compliant=18, NotApplicable=2, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 99999\n- password.min_age | Minimum password age | Non-Compliant | PASS_MIN_DAYS 0\n- password.complexity | Password complexity | Compliant | minclass=4, required classes=0\n- lockout.threshold | Account lockout threshold | Non-Compliant | pam_faillock not enabled\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 2 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Compliant | none\n- user.account_type | Signed-in user account type | Compliant | Standard (ewagner)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 1 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | dnf-automatic timer off, apply_updates=no\n- gdpr.updates_recent | Security updates installed recently | Compliant | 4 days ago\n- gdpr.os_supported | Supported operating system version | Non-Compliant | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Non-Compliant | none\n- gdpr.backup_recent | Recent backup | Non-Compliant | no backup\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | teamviewer, chromeremotedesktop\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Claude Code\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Claude Code\n- aiact.risk_classification | AI system risk classification | Pending | Claude Code\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | no\n- org.breach_procedure | Breach response procedure | Pending | not answered\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Non-Compliant | partial\n- org.ai_literacy | AI literacy training | Pending | not answered\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Non-Compliant | no\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.expiration: Low\n- password.min_age: Low\n- lockout.threshold: Medium\n- malware.service: High\n- malware.on_access: Medium\n- admin.approved: Medium\n- encryption.recovery: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.os_supported: High\n- gdpr.backup_configured: High\n- gdpr.backup_recent: Medium\n- gdpr.remote_access: High\n- aiact.unapproved: Medium\n- org.dpia: Medium\n- org.ai_register: Medium\n- org.ai_transparency: Low\nSCORE (computed): compliant=32, assessed=50, percent=64\nTOP GAPS (computed, highest risk first):\n- malware.service | ClamAV malware protection service | High | not running\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 1 critical/high\n- gdpr.auto_updates | Automatic security updates | High | dnf-automatic timer off, apply_updates=no\n- gdpr.os_supported | Supported operating system version | High | Red Hat Enterprise Linux 8.10 (Ootpa)\n- gdpr.backup_configured | Backup configured | High | none\nFIX GUIDANCE (highest-risk gaps):\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- vuln.known_vulnerabilities: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- gdpr.auto_updates: Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer."}, {"role": "assistant", "content": "{\"device\": \"legal-dev-07 - Red Hat Enterprise Linux 8.10 (Ootpa) 8.10\", \"score\": {\"compliant\": 32, \"assessed\": 50, \"percent\": 64}, \"counts\": {\"Compliant\": 32, \"Non-Compliant\": 18, \"NotApplicable\": 2, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"1 critical/high\"}, {\"check_id\": \"gdpr.auto_updates\", \"name\": \"Automatic security updates\", \"risk\": \"High\", \"found\": \"dnf-automatic timer off, apply_updates=no\"}, {\"check_id\": \"gdpr.os_supported\", \"name\": \"Supported operating system version\", \"risk\": \"High\", \"found\": \"Red Hat Enterprise Linux 8.10 (Ootpa)\"}, {\"check_id\": \"gdpr.backup_configured\", \"name\": \"Backup configured\", \"risk\": \"High\", \"found\": \"none\"}], \"summary\": \"legal-dev-07 meets 32 of 50 assessed checks (64%). The most serious gaps are ClamAV malware protection service, Known vulnerabilities in installed software and Automatic security updates. 4 items are pending.\", \"next_steps\": [\"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Known vulnerabilities in installed software: Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Automatic security updates: Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\", \"Answer the 2 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Found: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 2 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Compliant\", \"title\": \"Disk encryption recovery is compliant\", \"finding\": \"The device reports 2 key slots, 0 unlock tokens, which meets the requirement (recovery key or second unlock method).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 180 days\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Compliant\", \"title\": \"Password expiration: meets the baseline\", \"finding\": \"Found: 180 days. This satisfies the baseline requirement of 1-365 days.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001339, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 187 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-03-29 16:03:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security updates installed recently\", \"finding\": \"The device reports 187 days ago, but the baseline requires <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-EXYV7YY\\sali, DESKTOP-EXYV7YY\\Administrator\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts: does not meet the baseline\", \"finding\": \"Unauthorized employee local administrator accounts is DESKTOP-EXYV7YY\\\\sali, DESKTOP-EXYV7YY\\\\Administrator instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (windows):\n- Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\n- For a stand-alone device: net accounts /maxpwage:365\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"Found: never. Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set Group Policy: Account Policies > Password Policy > Maximum password age = 365 days.\", \"For a stand-alone device: net accounts /maxpwage:365\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Android 16 (android)\nAI TOOLS FOUND: ChatGPT, Grammarly, Microsoft Copilot\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- ChatGPT: Cloud AI assistant; provider OpenAI (US); data: Provider cloud outside the EEA unless an EU data residency plan is used; default risk minimal; GDPR: Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\n- Grammarly: AI writing assistant (keyboard and browser); provider Grammarly (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Reads everything typed where it is enabled, including personal data; use the business plan with a data processing agreement or disable it in work apps.\n- Microsoft Copilot: Cloud AI assistant; provider Microsoft (US); data: Microsoft cloud; enterprise plans keep data under Microsoft's commercial data protection terms; default risk minimal; GDPR: Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"ChatGPT\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Needs a business plan with a data processing agreement and a transfer basis (Art. 28, Art. 44); consumer accounts may use prompts for training unless opted out.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Grammarly\", \"kind\": \"AI writing assistant (keyboard and browser)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Reads everything typed where it is enabled, including personal data; use the business plan with a data processing agreement or disable it in work apps.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Microsoft Copilot\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Use the Microsoft 365 work account version covered by the organisation's Microsoft data processing terms; consumer Copilot is not covered.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"3 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: ChatGPT, Grammarly, Microsoft Copilot.\"}"}], "meta": {"task": "ai_classification", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] What is in the audit report: The report has a compliance score, the most important gaps, and for each problem: what was found, why it matters, the risk level, step-by-step fixes for that operating system, how to verify the fix, and the related ISO 27001, GDPR and EU AI Act references. It also lists AI tools found and their EU AI Act risk class.\n[2] Privacy: where do audit results go?: Nowhere. Evidence and reports stay on the device. The app downloads only two things: the AI auditor model and a signed list of known vulnerabilities. Nothing about the device is uploaded. Chat history is stored only in this browser.\n[3] Automatic security updates on macOS: Automatic security updates on macOS. Why it matters: Security fixes must be installed automatically so that known vulnerabilities are closed quickly. Risk if missing: High - Unpatched devices are the most common way attackers get in. Requirement: Security updates are downloaded and installed automatically. How to fix on macOS: (1) Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true). How to verify: Re-run the audit; Automatic security updates should be Compliant. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Lockout observation period on macOS: Lockout observation period on macOS. Why it matters: Failed attempts must be counted over a long enough window to detect slow guessing. Risk if missing: Low - A short window lets attackers pace their guesses to avoid lockout. Requirement: Failed-attempt counter kept for >= LOCKOUT_OBSERVATION_MIN_MINUTES minutes. How to fix on macOS: (1) On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile. How to verify: Re-run the audit; Lockout observation period should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: macOS 26.0.1 (macos), audited 2026-10-02; 36 of 49 assessed checks compliant; 13 gaps, 2 pending, 0 not read. Gaps: Lockout observation period: not set; Screen lock timeout: 1200 idle, password delay 0; USB storage device access: deny; USB storage read: deny; Unauthorized employee administrator accounts: ewagner; Disk encryption recovery: no recovery key; Known vulnerabilities in installed software: 69 critical/high; Security updates installed recently: 92 days ago.\n\nQUESTION: What is the most important thing to fix?"}, {"role": "assistant", "content": "Your audit found 13 gaps. Start with Lockout observation period (found: not set). Failed attempts must be counted over a long enough window to detect slow guessing. To fix it [4]:\n1. On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile.\n\nThen look at: Screen lock timeout, USB storage device access, USB storage read."}], "meta": {"task": "chat", "os": "macos", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Pending\", \"title\": \"AI transparency to users is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (linux):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users is non-compliant\", \"finding\": \"AI transparency to users is no instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: allow\nDETAIL: Not configured: Windows default (Allow) applies.\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Private outbound is non-compliant\", \"finding\": \"Found: allow. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (android):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000176, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 15 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Compliant\", \"title\": \"Account lockout duration matches the required setting\", \"finding\": \"The device reports 15 minutes, which meets the requirement (>= 15 minutes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Apple malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: XProtect scanning on; Gatekeeper assessments disabled\nREQUIRED: XProtect and Gatekeeper on\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (macos):\n- Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Apple malware protection service\", \"finding\": \"The device reports XProtect scanning on; Gatekeeper assessments disabled, but the baseline requires XProtect and Gatekeeper on. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"Re-enable Gatekeeper: sudo spctl --master-enable. Keep 'Install Security Responses and system files' on in Software Update.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"iCloud Backup is non-compliant\", \"finding\": \"iCloud Backup is no backup instead of backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Analytics sharing minimised on iPhone: Analytics sharing minimised on iPhone. Why it matters: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Risk if missing: Low - More personal data than necessary leaves the organisation. Requirement: OS diagnostic/usage data sharing is at the minimum level. How to fix on iPhone: (1) Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics. How to verify: Re-run the audit; Diagnostic data minimised should be Compliant. Related controls: ISO 27001:2022 A.5.34 (Privacy and protection of PII); GDPR Art. 5(1)(c) (Data minimisation); GDPR Art. 25 (Data protection by design and by default).\n[2] Where to find Automatic iOS updates on iPhone: Automatic iOS updates on iPhone: What do Automatic Updates show? Where to look: Settings › General › Software Update › Automatic Updates\n[3] Known vulnerabilities in iOS on iPhone: Known vulnerabilities in iOS on iPhone. Why it matters: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Risk if missing: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation. Requirement: Installed software versions matched against an offline copy of the OSV/NVD vulnerability data: critical or high vulnerabilities <= MAX_CRITICAL_HIGH_VULNERABILITIES. iOS: the iOS version is matched against NVD iOS/iPadOS records. How to fix on iPhone: (1) Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use. (2) Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it. How to verify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer. Related controls: ISO 27001:2022 A.8.8 (Management of technical vulnerabilities); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Where to find Analytics sharing minimised on iPhone: Analytics sharing minimised on iPhone: Is 'Share iPhone Analytics' on? Where to look: Settings › Privacy & Security › Analytics & Improvements\n\nMY AUDIT: none\n\nQUESTION: How do I see analytics sharing minimised on iPhone?"}, {"role": "assistant", "content": "Go to Settings › Privacy & Security › Analytics & Improvements [4]"}], "meta": {"task": "chat", "os": "ios", "status": "phone", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Android 16 (android)\nHOSTNAME: sales-xiaomi-35\nAUDIT RUN AS ADMIN: no\nCOUNTS: Compliant=22, Non-Compliant=4, NotApplicable=0, Pending=4, Error=0\nEVIDENCE: 3 detected by the browser, 19 self-reported by the user\nRESULTS (id | name | status | found):\n- malware.antivirus | Google Play Protect | Compliant | Play Protect scanning on\n- screen_lock.enabled | Screen lock | Compliant | screen lock set\n- screen_lock.timeout | Screen timeout | Compliant | Screen timeout 30 seconds\n- mobile.passcode_strength | Screen lock strength | Compliant | 6-digit PIN\n- mobile.jailbreak_root | Not rooted | Compliant | not rooted (user confirmed)\n- mobile.app_sources | Install unknown apps blocked | Compliant | unknown apps blocked for all apps\n- mobile.usb_debugging | USB debugging off | Compliant | off\n- mobile.find_my_device | Find My Device | Non-Compliant | off\n- encryption.enabled | Device encryption | Compliant | on (Android 16 encrypts by default)\n- vuln.pending_updates | Android security update | Compliant | security patch 2026-10 (0 months old)\n- vuln.outdated_apps | App updates | Compliant | automatic app updates on\n- gdpr.auto_updates | Automatic system updates | Compliant | automatic\n- gdpr.os_supported | Supported operating system version | Compliant | Android 16\n- gdpr.time_sync | Date & time set automatically | Pending | not answered\n- gdpr.backup_configured | Google backup | Compliant | Google backup on\n- gdpr.telemetry | Usage & diagnostics sharing minimised | Pending | not answered\n- gdpr.lock_screen_previews | Sensitive notifications hidden when locked | Compliant | no notifications on lock screen\n- aiact.inventory | AI software inventory | Compliant | ChatGPT\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | ChatGPT\n- aiact.risk_classification | AI system risk classification | Pending | ChatGPT\n- aiact.os_ai_features | AI assistant access restricted | Compliant | assistant screen access off\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Non-Compliant | partial\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | partial\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- mobile.find_my_device: Medium\n- aiact.unapproved: Medium\n- org.processor_agreements: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=22, assessed=26, percent=85\nTOP GAPS (computed, highest risk first):\n- org.processor_agreements | Data processing agreements | High | partial\n- mobile.find_my_device | Find My Device | Medium | off\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Medium | ChatGPT\n- org.ai_literacy | AI literacy training | Medium | partial\nFIX GUIDANCE (highest-risk gaps):\n- org.processor_agreements: List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\n- mobile.find_my_device: Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\n- aiact.unapproved: Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it."}, {"role": "assistant", "content": "{\"device\": \"sales-xiaomi-35 - Android 16\", \"score\": {\"compliant\": 22, \"assessed\": 26, \"percent\": 85}, \"counts\": {\"Compliant\": 22, \"Non-Compliant\": 4, \"NotApplicable\": 0, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"org.processor_agreements\", \"name\": \"Data processing agreements\", \"risk\": \"High\", \"found\": \"partial\"}, {\"check_id\": \"mobile.find_my_device\", \"name\": \"Find My Device\", \"risk\": \"Medium\", \"found\": \"off\"}, {\"check_id\": \"aiact.unapproved\", \"name\": \"Unapproved AI tools (shadow AI)\", \"risk\": \"Medium\", \"found\": \"ChatGPT\"}, {\"check_id\": \"org.ai_literacy\", \"name\": \"AI literacy training\", \"risk\": \"Medium\", \"found\": \"partial\"}], \"summary\": \"sales-xiaomi-35 meets 22 of 26 assessed checks (85%). The most serious gaps are Data processing agreements. 4 items are pending. 19 results are self-reported by the user and were not verified on the device.\", \"next_steps\": [\"Data processing agreements: List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\", \"Find My Device: Settings › Google › All services › Find My Device (Find Hub): turn on, and make sure Location is on.\", \"Unapproved AI tools (shadow AI): Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "android", "audit": 7001231, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Auto-Lock\", \"finding\": \"The user reports Auto-Lock Never, but the baseline requires <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Found: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000994, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Auto-Lock [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Auto-Lock Never\nREQUIRED: <= 300 seconds\nDETAIL: The phone never locks by itself.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (ios):\n- Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Auto-Lock\", \"finding\": \"Auto-Lock is Auto-Lock Never instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Settings › Display & Brightness › Auto-Lock: choose 5 minutes or less (not Never).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention is pending\", \"finding\": \"Required only if a tool is classified high-risk. Current state: depends on classification.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000641, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (ewagner)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (macos):\n- Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Signed-in user account type\", \"finding\": \"Found: Administrator (ewagner). Required: Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (System Settings > Users & Groups, or sudo dseditgroup -o edit -d <account> -t user admin) and keep a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Claude\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (macos):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"The device reports Claude, which meets the requirement (inventory taken).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"The device reports not running, but the baseline requires on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000689, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: all removable storage denied by policy\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (windows):\n- Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Non-Compliant\", \"title\": \"USB storage read: does not meet the baseline\", \"finding\": \"Found: deny. Required: allow. The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Administrative Templates > System > Removable Storage Access > Removable Disks: Deny read access = Disabled.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage read should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage read [usb.read] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: deny\nREQUIRED: allow\nDETAIL: usb-storage module disabled\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\nrisk: Low - The setting differs from the organisation's agreed USB policy.\nfix (linux):\n- To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or use USBGuard.\nverify: Re-run the audit; USB storage read should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.read\", \"status\": \"Non-Compliant\", \"title\": \"USB storage read is non-compliant\", \"finding\": \"USB storage read is deny instead of allow. The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter.\", \"risk\": {\"level\": \"Low\", \"reason\": \"The setting differs from the organisation's agreed USB policy.\"}, \"recommendation\": [\"To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or use USBGuard.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage read should read allow.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.read", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"The device reports no, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (linux):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000342, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Domain inbound\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001438, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: FileVault on\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Encryption matches the required setting\", \"finding\": \"The device reports FileVault on, which meets the requirement (FileVault on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"Gap: ClamAV malware protection service\", \"finding\": \"ClamAV malware protection service is not running instead of running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public inbound [firewall.public.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Public inbound is non-compliant\", \"finding\": \"Found: allow. Required: block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001427, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.public.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_LITERACY in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (macos):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_AI_LITERACY in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000075, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Not jailbroken [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: not jailbroken (user confirmed)\nREQUIRED: not jailbroken\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (ios):\n- Restore the iPhone with the latest iOS (Settings › General › Transfer or Reset iPhone › Erase All Content and Settings, or restore through a computer), then restore data from a backup.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Compliant\", \"title\": \"Not jailbroken is compliant\", \"finding\": \"Found: not jailbroken (user confirmed). This satisfies the baseline requirement of not jailbroken. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Recall feature Enabled, not disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Windows Recall / Copilot data analysis disabled\", \"finding\": \"Found: Recall feature Enabled, not disabled by policy. Required: disabled by policy. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signature updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: freshclam not running\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (linux):\n- systemctl enable --now clamav-freshclam.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware signature updates is non-compliant\", \"finding\": \"The device reports freshclam not running, but the baseline requires automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Gemini, Grok\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: Gemini, Grok. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: date not readable\nREQUIRED: <= 7 days ago\nDETAIL: Confirm the last backup date in: restic-backup.timer.\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (linux):\n- Run the backup job now and check its timer and logs.\nverify: Re-run the audit; the last backup should be within 7 days.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Pending\", \"title\": \"Recent backup: needs a follow-up step\", \"finding\": \"Current state: date not readable. Confirm the last backup date in: restic-backup.timer.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run the backup job now and check its timer and logs.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Ollama, Claude\nREQUIRED: inventory taken\nDETAIL: Local AI service listening on port 11434. Local models: /Users/jdoe/.ollama/models (59G).\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (macos):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"AI software inventory is set to Ollama, Claude, in line with the required inventory taken.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] Antivirus on Windows: Antivirus on Windows. Why it matters: Anti-malware software detects and blocks known malicious files and behaviour. Risk if missing: High - Without working anti-malware, ransomware and info-stealers can run unchallenged. Requirement: An enabled, up-to-date antivirus product is present. Android: Google Play Protect scanning is on. How to fix on Windows: (1) Turn Microsoft Defender Antivirus back on (Windows Security > Virus & threat protection) or make sure the third-party product is installed, enabled and reporting to Security Center. How to verify: Re-run the audit; Antivirus should be Compliant. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] Tamper protection on Windows: Tamper protection on Windows. Why it matters: Tamper protection stops malware or users switching off Defender. Risk if missing: Medium - Attackers commonly disable antivirus before deploying ransomware. Requirement: Defender tamper protection is on. How to fix on Windows: (1) Turn on Tamper Protection in Windows Security > Virus & threat protection settings, or centrally through Microsoft Intune / Defender for Endpoint (it cannot be set with PowerShell). How to verify: Re-run the audit; Tamper protection should be on. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] On-access protection on Windows: On-access protection on Windows. Why it matters: On-access scanning checks files when they are opened or written. Risk if missing: Medium - Malicious files can be opened before any scheduled scan finds them. Requirement: Files are scanned when opened. macOS: NotApplicable unless a third-party scanner provides it. How to fix on Windows: (1) Enable real-time protection in Microsoft Defender (on-access protection follows it). How to verify: Re-run the audit; On-access malware protection should be Compliant. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Microsoft Defender Antimalware on Windows: Microsoft Defender Antimalware on Windows. Why it matters: The malware protection service must be running for protection to work. Risk if missing: High - A stopped protection service leaves the device unprotected. Requirement: The platform malware protection service is running. How to fix on Windows: (1) Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device. How to verify: Re-run the audit; the malware protection service should be Compliant. Related controls: ISO 27001:2022 A.8.7 (Protection against malware); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: Windows 11 (windows) on a Windows PC, Edge 141; detected by the browser, security settings not checked.\n\nQUESTION: What's the risk if microsoft defender antimalware isn't set up on my Windows?"}, {"role": "assistant", "content": "The malware protection service must be running for protection to work. If it is missing, the risk is high: A stopped protection service leaves the device unprotected. [4]"}], "meta": {"task": "chat", "os": "windows", "status": "why", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private outbound [firewall.private.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Private outbound\", \"finding\": \"Found: block. Required: allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 14\nREQUIRED: >= 14\nDETAIL: pam_pwquality\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Compliant\", \"title\": \"Minimum password length: meets the baseline\", \"finding\": \"The device reports 14, which meets the requirement (>= 14).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (macos):\n- Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"The device reports none of the approved accounts is an administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 82 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last software update: 2026-07-15 20:17:00 +0000\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (macos):\n- Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Security updates installed recently: does not meet the baseline\", \"finding\": \"Security updates installed recently is 82 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Install pending updates (System Settings > General > Software Update, or softwareupdate -ia).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no signature database\nREQUIRED: <= 3 days old\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Non-Compliant\", \"title\": \"Malware signatures: does not meet the baseline\", \"finding\": \"Malware signatures is no signature database instead of <= 3 days old. Anti-malware can only detect threats it knows about; signatures must be current.\", \"risk\": {\"level\": \"High\", \"reason\": \"Outdated signatures miss recent malware.\"}, \"recommendation\": [\"systemctl enable --now clamav-freshclam, or run freshclam once to update now.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signatures should be within the allowed age.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"Gap: USB storage write\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: NTP service no, synchronised no\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Non-Compliant\", \"title\": \"Time synchronisation is non-compliant\", \"finding\": \"Found: NTP service no, synchronised no. Required: time synchronised. Accurate clocks are needed to put events in the right order during an investigation.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Wrong timestamps make incident timelines unreliable.\"}, \"recommendation\": [\"timedatectl set-ntp true (or enable chronyd).\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"], \"verification\": \"Re-run the audit; Time synchronisation should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no FileVault\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"Found: no FileVault. Required: recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: never\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (macos):\n- Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"Password expiration is never instead of 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Deploy a Passcode configuration profile with maxPINAgeInDays = 365.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (lmuller)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (linux):\n- Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type: does not meet the baseline\", \"finding\": \"Signed-in user account type is Administrator (lmuller) instead of Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured is non-compliant\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_BREACH_PROCEDURE in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (windows):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Answer ORG_BREACH_PROCEDURE in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 3 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Compliant\", \"title\": \"Disk encryption recovery is compliant\", \"finding\": \"Disk encryption recovery is set to 3 key slots, 0 unlock tokens, in line with the required recovery key or second unlock method.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Gemini, Grok\nREQUIRED: only: none\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (ios):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI): does not meet the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is Gemini, Grok instead of only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (macos):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7.1 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (ios):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Pending\", \"title\": \"Record of processing activities is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001520, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Lockout observation period [lockout.observation] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Failed attempts must be counted over a long enough window to detect slow guessing.\nrisk: Low - A short window lets attackers pace their guesses to avoid lockout.\nfix (linux):\n- Set fail_interval = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Lockout observation period should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.observation\", \"status\": \"Compliant\", \"title\": \"Lockout observation period is compliant\", \"finding\": \"The device reports 30 minutes, which meets the requirement (>= 15 minutes).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.observation", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: analytics sharing on\nREQUIRED: analytics sharing off\nDETAIL: Usage data from the phone is sent to Apple.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Analytics sharing minimised\", \"finding\": \"Found: analytics sharing on. Required: analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: AllowTelemetry=0 (Settings)\nREQUIRED: required diagnostic data only (0-1)\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Data Collection and Preview Builds > Allow Diagnostic Data = Send required diagnostic data (AllowTelemetry = 1; 0 on Enterprise/Education).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Compliant\", \"title\": \"Diagnostic data minimised: meets the baseline\", \"finding\": \"Found: AllowTelemetry=0 (Settings). This satisfies the baseline requirement of required diagnostic data only (0-1).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (windows):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"AI literacy training is non-compliant\", \"finding\": \"The device reports partial, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001083, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Encryption [encryption.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: FileVault on\nREQUIRED: FileVault on\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\nGUIDANCE:\nwhy: Full-disk encryption keeps data unreadable if the device is lost or stolen.\nrisk: High - A lost unencrypted laptop is a personal data breach that may have to be reported.\nfix (macos):\n- Turn on FileVault (System Settings > Privacy & Security > FileVault, or sudo fdesetup enable), ideally enforced and escrowed by your MDM.\nverify: Re-run the audit; Encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.enabled\", \"status\": \"Compliant\", \"title\": \"Encryption: meets the baseline\", \"finding\": \"The device reports FileVault on, which meets the requirement (FileVault on).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(a)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "encryption.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password history [password.history] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 24\nREQUIRED: >= 24\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Without a password history, users can return to old, possibly exposed passwords.\nrisk: Low - Reused passwords keep compromised credentials valid.\nfix (macos):\n- Deploy a Passcode configuration profile with pinHistory = 24.\nverify: Re-run the audit; Password history should read 24 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.history\", \"status\": \"Compliant\", \"title\": \"Password history is compliant\", \"finding\": \"Found: 24. This satisfies the baseline requirement of >= 24.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000603, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "password.history", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow (no firewall)\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (linux):\n- ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy is non-compliant\", \"finding\": \"Found: allow (no firewall). Required: deny. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"ufw default deny incoming (firewalld: keep the default zone target at default/REJECT).\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: LightDM guest session allowed\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Guest account disabled\", \"finding\": \"The device reports LightDM guest session allowed, but the baseline requires no guest login. Guest logins let people use the device without an identifiable account.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Actions by a guest cannot be traced to a person, and guests may reach shared data.\"}, \"recommendation\": [\"Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Guest account disabled should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001290, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: share Mac analytics=1, share with app developers=0\nREQUIRED: analytics sharing off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (macos):\n- Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised: does not meet the baseline\", \"finding\": \"The device reports share Mac analytics=1, share with app developers=0, but the baseline requires analytics sharing off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Turn off System Settings > Privacy & Security > Analytics & Improvements > Share Mac Analytics and Share with App Developers, or enforce by profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: iPadOS 26.7 (ios)\nAI TOOLS FOUND: DeepSeek, Meta AI\nAPPROVED AI TOOLS: none\nINVENTORY DETAIL: AI apps listed by the user.\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- DeepSeek: Cloud AI assistant; provider DeepSeek (China); data: Provider cloud in the People's Republic of China; default risk minimal; GDPR: No EU adequacy decision for China and no business data processing terms; EU data protection authorities have restricted or investigated it. Do not use with work or personal data.\n- Meta AI: Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps); provider Meta (US); data: Provider cloud outside the EEA; default risk minimal; GDPR: Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"DeepSeek\", \"kind\": \"Cloud AI assistant\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"No EU adequacy decision for China and no business data processing terms; EU data protection authorities have restricted or investigated it. Do not use with work or personal data.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}, {\"tool\": \"Meta AI\", \"kind\": \"Cloud AI assistant (built into WhatsApp, Instagram and Facebook apps)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Consumer service with no business data processing terms; do not share work or customer data in chats with Meta AI.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"2 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: DeepSeek, Meta AI.\"}"}], "meta": {"task": "ai_classification", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware signatures [malware.signatures] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 2 days old\nREQUIRED: <= 3 days old\nDETAIL: /var/lib/clamav/daily.cld\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware can only detect threats it knows about; signatures must be current.\nrisk: High - Outdated signatures miss recent malware.\nfix (linux):\n- systemctl enable --now clamav-freshclam, or run freshclam once to update now.\nverify: Re-run the audit; signatures should be within the allowed age."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signatures\", \"status\": \"Compliant\", \"title\": \"Malware signatures matches the required setting\", \"finding\": \"The device reports 2 days old, which meets the requirement (<= 3 days old).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.signatures", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: ConfigDataInstall=0, CriticalUpdateInstall=1\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Non-Compliant\", \"title\": \"Malware security updates: does not meet the baseline\", \"finding\": \"The device reports ConfigDataInstall=0, CriticalUpdateInstall=1, but the baseline requires automatic. Signature updates must happen automatically so protection stays current without user action.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Manual updates are forgotten, leaving protection outdated.\"}, \"recommendation\": [\"Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; signature updates should be automatic.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (linux):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Compliant\", \"title\": \"Human oversight for high-risk AI matches the required setting\", \"finding\": \"Human oversight for high-risk AI is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000146, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nDETAIL: No configuration profile restricts external drives.\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write is non-compliant\", \"finding\": \"Found: allow. Required: deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000335, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised is non-compliant\", \"finding\": \"Usage & diagnostics sharing minimised is usage & diagnostics on instead of usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000511, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: dnf-automatic timer off, apply_updates=no\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (linux):\n- Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates: does not meet the baseline\", \"finding\": \"Automatic security updates is dnf-automatic timer off, apply_updates=no instead of automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Debian/Ubuntu: apt install unattended-upgrades && dpkg-reconfigure -plow unattended-upgrades. RHEL: dnf install dnf-automatic && systemctl enable --now dnf-automatic-install.timer.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: 30 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Compliant\", \"title\": \"Account lockout duration matches the required setting\", \"finding\": \"Account lockout duration is set to 30 minutes, in line with the required >= 15 minutes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (android):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not set\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration: does not meet the baseline\", \"finding\": \"Account lockout duration is not set instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000524, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (ios):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Pending\", \"title\": \"Breach response procedure is pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000553, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: usage & diagnostics on\nREQUIRED: usage & diagnostics off\nDETAIL: Usage data from the phone is sent to Google.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Usage & diagnostics sharing minimised: does not meet the baseline\", \"finding\": \"Usage & diagnostics sharing minimised is usage & diagnostics on instead of usage & diagnostics off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000978, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none of the approved accounts is an administrator\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (macos):\n- Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Non-Compliant\", \"title\": \"Approved IT administrator: does not meet the baseline\", \"finding\": \"The device reports none of the approved accounts is an administrator, but the baseline requires itadmin. A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\"}, \"recommendation\": [\"Create or promote the approved IT account: sudo dseditgroup -o edit -a <account> -t user admin, ideally managed by your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Approved IT administrator should list the approved account.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000256, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Pending security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 pending\nREQUIRED: <= 0 pending\nDETAIL: Including: libssl3t64, linux-image-generic, libc6. Package lists are 14 days old.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (linux):\n- Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Pending security updates\", \"finding\": \"The device reports 15 pending, but the baseline requires <= 0 pending. Security updates that are offered but not installed leave known, published vulnerabilities open.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: screen saver does not require sign-in for: DESKTOP-J0Q0Y2I\\ewagner\nREQUIRED: sign-in required to unlock\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (windows):\n- Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"Found: screen saver does not require sign-in for: DESKTOP-J0Q0Y2I\\\\ewagner. Required: sign-in required to unlock. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"Set the machine inactivity limit by Group Policy (this always requires sign-in), or enable 'Password protect the screen saver' for users by Group Policy.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rschmit\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (linux):\n- Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee administrator accounts: does not meet the baseline\", \"finding\": \"Found: rschmit. Required: none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (windows):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory matches the required setting\", \"finding\": \"The device reports no AI tools found, which meets the requirement (inventory taken).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000975, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB drives writable without encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (windows):\n- Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Removable media encryption\", \"finding\": \"Found: USB drives writable without encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Enable 'Deny write access to removable drives not protected by BitLocker' (Group Policy: Windows Components > BitLocker Drive Encryption > Removable Data Drives), or deny USB write access.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"Gap: On-access malware protection\", \"finding\": \"Found: not running. Required: on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: DESKTOP-4215OE9\\asmith\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts is non-compliant\", \"finding\": \"Unauthorized employee local administrator accounts is DESKTOP-4215OE9\\\\asmith instead of none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000969, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Not rooted [mobile.jailbreak_root] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: rooted or bootloader unlocked\nREQUIRED: not rooted\nDETAIL: Android security protections are weakened on a rooted phone.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Jailbreaking or rooting removes the operating system's app isolation and security checks.\nrisk: High - Any app can then read other apps' data, including work email and files.\nfix (android):\n- Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\nverify: Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.jailbreak_root\", \"status\": \"Non-Compliant\", \"title\": \"Not rooted is non-compliant\", \"finding\": \"Found: rooted or bootloader unlocked. Required: not rooted. Jailbreaking or rooting removes the operating system's app isolation and security checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Any app can then read other apps' data, including work email and files.\"}, \"recommendation\": [\"Remove root and re-lock the bootloader by installing the phone maker's official firmware, or stop using the phone for work.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check after restoring; the phone should be reported as not jailbroken / not rooted.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000931, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "mobile.jailbreak_root", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: Recall not available on this device\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"NotApplicable\", \"title\": \"Windows Recall / Copilot data analysis disabled: not assessed on this device\", \"finding\": \"Not applicable on this device: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 45 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 45\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"The device reports 45 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000827, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 142 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iOS 26.4 checked against NVD iOS records. iOS 26.4 (142 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-28846). 208 medium/low. Patch-level versions (x.y.z) are not always visible to the browser; confirm the version in Settings › General › About.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS is non-compliant\", \"finding\": \"Known vulnerabilities in iOS is 142 critical/high instead of <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001240, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (linux):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: decision pending\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000962, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: decision pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000901, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Recent backup [gdpr.backup_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: <= 7 days ago\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A configured backup only helps if it has actually run recently.\nrisk: Medium - Recent work would be lost if the device failed today.\nfix (macos):\n- Run tmutil startbackup and check that the destination is reachable.\nverify: Re-run the audit; the last backup should be within 7 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_recent\", \"status\": \"Non-Compliant\", \"title\": \"Recent backup: does not meet the baseline\", \"finding\": \"Recent backup is no backup instead of <= 7 days ago. A configured backup only helps if it has actually run recently.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Recent work would be lost if the device failed today.\"}, \"recommendation\": [\"Run tmutil startbackup and check that the destination is reachable.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; the last backup should be within 7 days.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000791, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Private inbound [firewall.private.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.private.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Private inbound: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.private.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (ios):\n- Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"App updates is non-compliant\", \"finding\": \"App updates is automatic app updates off instead of automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Settings › App Store: turn on App Updates; then App Store › account picture › update all pending apps.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: ChatGPT, Claude\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification: decision pending\", \"finding\": \"Current state: ChatGPT, Claude. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001180, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Malware security updates [malware.signature_updates] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: security responses and system data files install automatically\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Signature updates must happen automatically so protection stays current without user action.\nrisk: Medium - Manual updates are forgotten, leaving protection outdated.\nfix (macos):\n- Turn on 'Install Security Responses and system files' in System Settings > General > Software Update > Automatic Updates, or set ConfigDataInstall and CriticalUpdateInstall to true by configuration profile.\nverify: Re-run the audit; signature updates should be automatic."}, {"role": "assistant", "content": "{\"check_id\": \"malware.signature_updates\", \"status\": \"Compliant\", \"title\": \"Malware security updates: meets the baseline\", \"finding\": \"Malware security updates is set to security responses and system data files install automatically, in line with the required automatic.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.signature_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (macos):\n- Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Backup configured: does not meet the baseline\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Configure Time Machine (tmutil setdestination) or the organisation's backup agent.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000129, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (ios):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Current state: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001418, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 16 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_PROCESSOR_AGREEMENTS in organisational.conf.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000040, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nHOSTNAME: ops-lt-51\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=28, Non-Compliant=17, NotApplicable=7, Pending=3, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 90 days\n- password.min_age | Minimum password age | Non-Compliant | PASS_MIN_DAYS 0\n- password.complexity | Password complexity | Compliant | minclass=3, required classes=0\n- lockout.threshold | Account lockout threshold | Non-Compliant | 10\n- lockout.duration | Account lockout duration | Non-Compliant | 10 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | 10 minutes\n- malware.antivirus | Antivirus | Compliant | CrowdStrike Falcon\n- malware.service | ClamAV malware protection service | NotApplicable | not used\n- malware.signature_updates | Malware signature updates | NotApplicable | not used\n- malware.signatures | Malware signatures | NotApplicable | not used\n- malware.on_access | On-access malware protection | Error | not read\n- firewall.enabled | Firewall | Compliant | firewalld active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: rschmit (1800s)\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Non-Compliant | none of the approved accounts is an administrator\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | jdoe\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (rschmit)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Non-Compliant | 1 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 37 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 11 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 1930 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | dnf-automatic timer on, apply_updates=yes\n- gdpr.updates_recent | Security updates installed recently | Compliant | 19 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Red Hat Enterprise Linux 9.6 (Plow)\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal volatile\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Deja Dup (jdoe)\n- gdpr.backup_recent | Recent backup | Non-Compliant | 57 days ago\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to USB storage blocked\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | anydesk\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Pending | not answered\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.min_age: Low\n- lockout.threshold: Medium\n- lockout.duration: Low\n- lockout.observation: Low\n- screen_lock.timeout: Medium\n- admin.approved: Medium\n- admin.unauthorized: High\n- user.account_type: High\n- encryption.recovery: Medium\n- vuln.pending_updates: High\n- vuln.outdated_apps: Medium\n- vuln.known_vulnerabilities: High\n- gdpr.backup_recent: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- org.ropa: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=28, assessed=45, percent=62\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee administrator accounts | High | jdoe\n- user.account_type | Signed-in user account type | High | Administrator (rschmit)\n- vuln.pending_updates | Pending security updates | High | 37 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 1930 critical/high\n- gdpr.remote_access | Remote access services restricted | High | anydesk\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- user.account_type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\n- vuln.pending_updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed."}, {"role": "assistant", "content": "{\"device\": \"ops-lt-51 - Red Hat Enterprise Linux 9.6 (Plow) 9.6\", \"score\": {\"compliant\": 28, \"assessed\": 45, \"percent\": 62}, \"counts\": {\"Compliant\": 28, \"Non-Compliant\": 17, \"NotApplicable\": 7, \"Pending\": 3, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"jdoe\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (rschmit)\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"37 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"1930 critical/high\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"anydesk\"}], \"summary\": \"ops-lt-51 meets 28 of 45 assessed checks (62%). The most serious gaps are Unauthorized employee administrator accounts, Signed-in user account type and Pending security updates. 1 checks could not be read. 3 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Signed-in user account type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\", \"Pending security updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\", \"Answer the 3 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000357, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (windows):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Data processing agreements is non-compliant\", \"finding\": \"Data processing agreements is no instead of yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Automatic iOS updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: download and install on\nREQUIRED: download and install on\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (ios):\n- Settings › General › Software Update › Automatic Updates: turn on iOS Updates Download and Install, and Security Responses & System Files.\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Compliant\", \"title\": \"Automatic iOS updates: meets the baseline\", \"finding\": \"Automatic iOS updates is set to download and install on, in line with the required download and install on. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: no\nCHECK: ClamAV malware protection service [malware.service] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (linux):\n- systemctl enable --now clamav-daemon (RHEL: clamd@scan).\nverify: Re-run the audit; the malware protection service should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"Non-Compliant\", \"title\": \"ClamAV malware protection service: does not meet the baseline\", \"finding\": \"Found: not running. Required: running. The malware protection service must be running for protection to work.\", \"risk\": {\"level\": \"High\", \"reason\": \"A stopped protection service leaves the device unprotected.\"}, \"recommendation\": [\"systemctl enable --now clamav-daemon (RHEL: clamd@scan).\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the malware protection service should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000516, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: Recall disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=1; AllowRecallEnablement=0; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Compliant\", \"title\": \"Windows Recall / Copilot data analysis disabled is compliant\", \"finding\": \"Windows Recall / Copilot data analysis disabled is set to Recall disabled by policy, in line with the required disabled by policy.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000616, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: Never\nREQUIRED: When Unlocked or Never\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Compliant\", \"title\": \"Notification previews hidden when locked matches the required setting\", \"finding\": \"Notification previews hidden when locked is set to Never, in line with the required When Unlocked or Never. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000800, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Account lockout duration\", \"finding\": \"The device reports 5 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 11 outdated\nREQUIRED: <= 0 outdated\nDETAIL: libpng (4.4.8) < 4.5.12; xz (9.18.2) < 9.20.17; readline (8.16.8) < 8.18.18; openssl@3 (7.20.8) < 7.21.17; wget (30.20.9) < 30.20.18.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (macos):\n- Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications: does not meet the baseline\", \"finding\": \"Outdated third-party applications is 11 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications from the App Store or their own updaters (or brew upgrade for Homebrew apps), and manage updates centrally with your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Breach response procedure [org.breach_procedure] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.24 Information security incident management planning and preparation: Plan in advance how security incidents are reported, handled and communicated.\n- ISO 27001:2022 A.5.26 Response to information security incidents: Respond to incidents according to the documented procedure.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Breaches must be reported to the supervisory authority within 72 hours where required.\nrisk: High - Without a tested procedure, the 72-hour deadline is easily missed.\nfix (linux):\n- Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\nverify: Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested."}, {"role": "assistant", "content": "{\"check_id\": \"org.breach_procedure\", \"status\": \"Non-Compliant\", \"title\": \"Breach response procedure is non-compliant\", \"finding\": \"Found: partial. Required: yes. Breaches must be reported to the supervisory authority within 72 hours where required.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without a tested procedure, the 72-hour deadline is easily missed.\"}, \"recommendation\": [\"Write a breach procedure (detection, assessment, 72-hour notification to the CNPD or your authority, informing people, breach log) and test it with an exercise.\"], \"references\": [\"ISO 27001:2022 A.5.24\", \"ISO 27001:2022 A.5.26\", \"GDPR Art. 33\", \"GDPR Art. 34\"], \"verification\": \"Answer ORG_BREACH_PROCEDURE=yes once the procedure is written and tested.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000494, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.breach_procedure", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (ios):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI transparency to users\", \"finding\": \"The device reports no, but the baseline requires yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Automatic security updates [gdpr.auto_updates] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: check=1 download=1 security=1 macOS-install=0\nREQUIRED: automatic\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\nrisk: High - Unpatched devices are the most common way attackers get in.\nfix (macos):\n- Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\nverify: Re-run the audit; Automatic security updates should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.auto_updates\", \"status\": \"Non-Compliant\", \"title\": \"Automatic security updates is non-compliant\", \"finding\": \"The device reports check=1 download=1 security=1 macOS-install=0, but the baseline requires automatic. Security fixes must be installed automatically so that known vulnerabilities are closed quickly.\", \"risk\": {\"level\": \"High\", \"reason\": \"Unpatched devices are the most common way attackers get in.\"}, \"recommendation\": [\"Turn on all options in System Settings > General > Software Update > Automatic Updates, or enforce them with a configuration profile (AutomaticallyInstallMacOSUpdates = true).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Automatic security updates should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000845, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.auto_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (macos):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Current state: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001500, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (android):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register: decision pending\", \"finding\": \"Answer ORG_AI_REGISTER in organisational.conf. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (ios):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training is compliant\", \"finding\": \"AI literacy training is set to yes, in line with the required yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000526, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 10 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Install unknown apps blocked: does not meet the baseline\", \"finding\": \"Found: unknown apps allowed for at least one app. Required: all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000269, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength: does not meet the baseline\", \"finding\": \"Found: 4-digit code. Required: >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000622, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.8 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store: does not meet the baseline\", \"finding\": \"The user reports apps from outside the App Store, but the baseline requires App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001417, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\n- For a stand-alone device: net accounts /minpwlen:14\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Minimum password length: does not meet the baseline\", \"finding\": \"Minimum password length is 6 instead of >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Account Policies > Password Policy > Minimum password length = 14 (or the Intune equivalent).\", \"For a stand-alone device: net accounts /minpwlen:14\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nHOSTNAME: legal-ws-55\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=28, Non-Compliant=22, NotApplicable=4, Pending=1, Error=1\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Non-Compliant | 6\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 90; accounts: jdoe (99999)\n- password.min_age | Minimum password age | Non-Compliant | PASS_MIN_DAYS 0\n- password.complexity | Password complexity | Non-Compliant | pam_pwquality not enabled\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | 15 minutes\n- lockout.observation | Lockout observation period | Non-Compliant | not set\n- malware.antivirus | Antivirus | Non-Compliant | none running\n- malware.service | ClamAV malware protection service | Non-Compliant | not running\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 2 days old\n- malware.on_access | On-access malware protection | Non-Compliant | not running\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Non-Compliant | off for: asmith\n- screen_lock.timeout | Screen lock timeout | Non-Compliant | too long for: asmith (1800s)\n- screen_lock.password | Screen lock password requirement | Non-Compliant | not required for: asmith\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Error | not read\n- user.account_type | Signed-in user account type | Non-Compliant | Administrator (asmith)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Compliant | 0 pending\n- vuln.outdated_apps | Outdated third-party applications | Non-Compliant | 40 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Non-Compliant | 202 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 22.04.5 LTS\n- gdpr.security_logging | Security event logging | Non-Compliant | auditd not running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | Timeshift\n- gdpr.backup_recent | Recent backup | Non-Compliant | 82 days ago\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Compliant | usage reporting off\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | no AI tools found\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Compliant | none\n- aiact.risk_classification | AI system risk classification | NotApplicable | no AI tools found\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | NotApplicable | no AI tools found\n- org.ropa | Record of processing activities | Non-Compliant | partial\n- org.dpia | Data protection impact assessment | Compliant | yes\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Non-Compliant | no\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Pending | not answered\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.min_length: Medium\n- password.expiration: Low\n- password.min_age: Low\n- password.complexity: Medium\n- lockout.observation: Low\n- malware.antivirus: High\n- malware.service: High\n- malware.on_access: Medium\n- screen_lock.enabled: Medium\n- screen_lock.timeout: Medium\n- screen_lock.password: Medium\n- usb.write: Medium\n- user.account_type: High\n- vuln.outdated_apps: Medium\n- gdpr.updates_recent: High\n- gdpr.security_logging: Medium\n- gdpr.backup_recent: Medium\n- gdpr.removable_encryption: Medium\n- org.ropa: High\n- org.breach_procedure: High\n- org.processor_agreements: High\n- org.ai_literacy: Medium\nSCORE (computed): compliant=28, assessed=50, percent=56\nTOP GAPS (computed, highest risk first):\n- malware.antivirus | Antivirus | High | none running\n- malware.service | ClamAV malware protection service | High | not running\n- user.account_type | Signed-in user account type | High | Administrator (asmith)\n- gdpr.updates_recent | Security updates installed recently | High | 202 days ago\n- org.ropa | Record of processing activities | High | partial\nFIX GUIDANCE (highest-risk gaps):\n- malware.antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\n- malware.service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\n- user.account_type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account."}, {"role": "assistant", "content": "{\"device\": \"legal-ws-55 - Ubuntu 22.04.5 LTS 22.04\", \"score\": {\"compliant\": 28, \"assessed\": 50, \"percent\": 56}, \"counts\": {\"Compliant\": 28, \"Non-Compliant\": 22, \"NotApplicable\": 4, \"Pending\": 1, \"Error\": 1}, \"top_gaps\": [{\"check_id\": \"malware.antivirus\", \"name\": \"Antivirus\", \"risk\": \"High\", \"found\": \"none running\"}, {\"check_id\": \"malware.service\", \"name\": \"ClamAV malware protection service\", \"risk\": \"High\", \"found\": \"not running\"}, {\"check_id\": \"user.account_type\", \"name\": \"Signed-in user account type\", \"risk\": \"High\", \"found\": \"Administrator (asmith)\"}, {\"check_id\": \"gdpr.updates_recent\", \"name\": \"Security updates installed recently\", \"risk\": \"High\", \"found\": \"202 days ago\"}, {\"check_id\": \"org.ropa\", \"name\": \"Record of processing activities\", \"risk\": \"High\", \"found\": \"partial\"}], \"summary\": \"legal-ws-55 meets 28 of 50 assessed checks (56%). The most serious gaps are Antivirus, ClamAV malware protection service and Signed-in user account type. 1 checks could not be read. 1 items are pending.\", \"next_steps\": [\"Re-run the audit as administrator/root so that every check can be read.\", \"Antivirus: Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\", \"ClamAV malware protection service: systemctl enable --now clamav-daemon (RHEL: clamd@scan).\", \"Signed-in user account type: Remove the signed-in user from sudo/wheel (gpasswd -d <account> sudo) and use a separate approved admin account.\", \"Answer the 1 open organisational questions in organisational.conf.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no lockout configured\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"Account lockout duration is no lockout configured instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000927, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4 outdated\nREQUIRED: <= 0 outdated\nDETAIL: Microsoft Visual Studio Code (User) Microsoft.VisualStudioCode 24.8.979 24.0.889 winget; Git Git.Git 11.7.268 11.3.115 winget; Notepad++ (64-bit x64) Notepad++.Notepad++ 8.6.610 8.7.444 winget; Mozilla Firefox (x64 en-US) Mozilla.Firefox 47.8.796 47.0.427 winget.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (windows):\n- Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Outdated third-party applications\", \"finding\": \"Outdated third-party applications is 4 outdated instead of <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update the listed applications (winget upgrade --all, or each app's own updater) and manage updates centrally with Intune or a patch management tool.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (linux):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Non-Compliant\", \"title\": \"Gap: AI literacy training\", \"finding\": \"The device reports no, but the baseline requires yes. Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Install unknown apps blocked [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: unknown apps allowed for at least one app\nREQUIRED: all apps: Not allowed\nDETAIL: Apps from outside Google Play skip Play Store checks.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (android):\n- Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Install unknown apps blocked\", \"finding\": \"Install unknown apps blocked is unknown apps allowed for at least one app instead of all apps: Not allowed. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Settings › Apps › Special app access › Install unknown apps: set every app to Not allowed; uninstall apps that did not come from Google Play.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001206, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Outdated third-party applications [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 42 outdated\nREQUIRED: <= 0 outdated\nDETAIL: packages: 42\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (linux):\n- Update packages and app stores: apt upgrade, snap refresh, flatpak update.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Outdated third-party applications is non-compliant\", \"finding\": \"The device reports 42 outdated, but the baseline requires <= 0 outdated. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Update packages and app stores: apt upgrade, snap refresh, flatpak update.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000948, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: allow\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (linux):\n- Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Non-Compliant\", \"title\": \"USB storage write: does not meet the baseline\", \"finding\": \"The device reports allow, but the baseline requires deny. Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\"}, \"recommendation\": [\"Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; USB storage write should read deny.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001275, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Guest account disabled [gdpr.guest_account] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: no guest login\nREQUIRED: no guest login\nREFERENCES:\n- ISO 27001:2022 A.5.16 Identity management: Every account belongs to an identifiable person or a documented service, through its whole life cycle.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Guest logins let people use the device without an identifiable account.\nrisk: Medium - Actions by a guest cannot be traced to a person, and guests may reach shared data.\nfix (linux):\n- Set allow-guest=false in the LightDM configuration and remove any 'guest' account with a login shell.\nverify: Re-run the audit; Guest account disabled should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.guest_account\", \"status\": \"Compliant\", \"title\": \"Guest account disabled is compliant\", \"finding\": \"Guest account disabled is set to no guest login, in line with the required no guest login.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.16\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.guest_account", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_AI_REGISTER in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (windows):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Pending\", \"title\": \"AI register is pending\", \"finding\": \"Current state: not answered. Answer ORG_AI_REGISTER in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: none running\nREQUIRED: antivirus running\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (linux):\n- Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Antivirus\", \"finding\": \"The device reports none running, but the baseline requires antivirus running. Anti-malware software detects and blocks known malicious files and behaviour.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Install an endpoint protection product, or ClamAV: apt install clamav clamav-daemon (RHEL: dnf install clamav clamd), then systemctl enable --now clamav-daemon.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001380, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode [screen_lock.enabled] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: passcode set\nREQUIRED: passcode set\nDETAIL: The browser reports that a device unlock method (passcode, PIN, Face ID or fingerprint) is set up.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An idle device should lock itself so nobody can use it while the owner is away.\nrisk: Medium - An unlocked, unattended device exposes all the user's data and accounts.\nfix (ios):\n- Settings › Face ID & Passcode › Turn Passcode On; for work phones the employer can require it with a configuration profile.\nverify: Re-run the audit; Screen lock should be on."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.enabled\", \"status\": \"Compliant\", \"title\": \"Passcode is compliant\", \"finding\": \"The device reports passcode set, which meets the requirement (passcode set).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001066, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.enabled", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: Character.AI, ChatGPT, Gemini\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (android):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Current state: Character.AI, ChatGPT, Gemini. Classified by the auditor model (Caveat classifier) in a later step. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000278, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Claude Code, ChatGPT, Microsoft Copilot\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Non-Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is non-compliant\", \"finding\": \"Found: Claude Code, ChatGPT, Microsoft Copilot. Required: only: none. AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\"}, \"recommendation\": [\"Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\", \"Publish an AI acceptable-use policy so staff know which tools are allowed.\", \"Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; Unapproved AI tools should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000267, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: USB storage write [usb.write] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: deny\nREQUIRED: deny\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Writing to USB drives is a common way for data to leave the organisation, often unencrypted.\nrisk: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost.\nfix (macos):\n- Restrict external storage to read-only or deny it with your MDM or an endpoint DLP product.\nverify: Re-run the audit; USB storage write should read deny."}, {"role": "assistant", "content": "{\"check_id\": \"usb.write\", \"status\": \"Compliant\", \"title\": \"USB storage write is compliant\", \"finding\": \"The device reports deny, which meets the requirement (deny).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001293, "profile": "typical", "dataset_version": "0.5.0", "check_id": "usb.write", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 1 key slots, 0 unlock tokens\nREQUIRED: recovery key or second unlock method\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (linux):\n- Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"The device reports 1 key slots, 0 unlock tokens, but the baseline requires recovery key or second unlock method. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Add a second key slot holding a recovery passphrase: cryptsetup luksAddKey <device>, and store it in your password vault.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000117, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: XProtect only\nREQUIRED: on-access scanning\nDETAIL: XProtect scans at launch and download; macOS has no built-in on-access scanning.\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (macos):\n- Not provided by macOS itself; deploy a managed endpoint protection product if required.\nverify: Re-run the audit; On-access malware protection should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"NotApplicable\", \"title\": \"On-access malware protection: not assessed on this device\", \"finding\": \"Not applicable on this device: XProtect scans at launch and download; macOS has no built-in on-access scanning.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Antivirus [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: XProtect 5396\nREQUIRED: antivirus present\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (macos):\n- XProtect is built in; if it is missing or damaged, reinstall the latest macOS update. Add a managed endpoint protection product if your policy requires on-access scanning.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Compliant\", \"title\": \"Antivirus: meets the baseline\", \"finding\": \"Found: XProtect 5396. This satisfies the baseline requirement of antivirus present.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001414, "profile": "typical", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 17.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Supported operating system version [gdpr.os_supported] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: iOS 17.7\nREQUIRED: >= iOS 18\nDETAIL: Version reported by the browser (Safari 17.7). This version no longer receives all security updates.\nEVIDENCE: detected by the browser\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: An operating system that no longer receives security updates cannot be kept secure.\nrisk: High - New vulnerabilities in an unsupported OS are never fixed.\nfix (ios):\n- Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\nverify: Re-run the audit; Supported operating system version should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.os_supported\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Supported operating system version\", \"finding\": \"Found: iOS 17.7. Required: >= iOS 18. An operating system that no longer receives security updates cannot be kept secure.\", \"risk\": {\"level\": \"High\", \"reason\": \"New vulnerabilities in an unsupported OS are never fixed.\"}, \"recommendation\": [\"Update to iOS 18 or later in Settings › General › Software Update; if the iPhone cannot run it, replace the iPhone.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Supported operating system version should be Compliant.\", \"evidence\": \"detected\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001441, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.os_supported", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 90; accounts: jdoe (99999)\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"Password expiration is PASS_MAX_DAYS 90; accounts: jdoe (99999) instead of 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (windows):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification: not applicable\", \"finding\": \"AI system risk classification is not assessed here: no AI tools found.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 13 (trixie) 13 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Minimum password length [password.min_length] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 6\nREQUIRED: >= 14\nDETAIL: pam_pwquality not enabled; pam_unix minimum\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\nrisk: Medium - Weak passwords make account takeover through guessing or credential stuffing easier.\nfix (linux):\n- Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\n- Set minlen = 14 in /etc/security/pwquality.conf.\nverify: Re-run the audit; Minimum password length should read 14 or more."}, {"role": "assistant", "content": "{\"check_id\": \"password.min_length\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Minimum password length\", \"finding\": \"Found: 6. Required: >= 14. Short passwords can be guessed or cracked quickly; length is the strongest single factor in password strength.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Weak passwords make account takeover through guessing or credential stuffing easier.\"}, \"recommendation\": [\"Install pam_pwquality (Debian/Ubuntu: apt install libpam-pwquality) and make sure it is listed in the PAM password stack.\", \"Set minlen = 14 in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Minimum password length should read 14 or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001063, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.min_length", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment: does not meet the baseline\", \"finding\": \"Data protection impact assessment is no instead of yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 13 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Pending\", \"title\": \"AI literacy training: needs a follow-up step\", \"finding\": \"Found: not answered. Answered once by the organisation (organisational.conf); not part of the phone self-check.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Untrained staff misuse AI tools and share data they should not.\"}, \"recommendation\": [\"Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"], \"verification\": \"Answer ORG_AI_LITERACY=yes once training is delivered and recorded.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000183, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iCloud Backup [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no backup\nREQUIRED: backup on\nDETAIL: Data on a lost or broken phone cannot be restored.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (ios):\n- Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iCloud Backup\", \"finding\": \"Found: no backup. Required: backup on. Backups make it possible to restore personal data after loss, theft, failure or ransomware. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Settings › [your name] › iCloud › iCloud Backup: turn on, or use the employer's approved backup; keep work data only in company-approved apps.\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000779, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Windows Recall / Copilot data analysis disabled [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: Recall feature Enabled, not disabled by policy\nREQUIRED: disabled by policy\nDETAIL: DisableAIDataAnalysis=; AllowRecallEnablement=; TurnOffWindowsCopilot=\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (windows):\n- Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Windows Recall / Copilot data analysis disabled\", \"finding\": \"Found: Recall feature Enabled, not disabled by policy. Required: disabled by policy. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Set Group Policy: Administrative Templates > Windows Components > Windows AI > Turn off saving snapshots for use with Recall = Enabled (DisableAIDataAnalysis = 1), unless Recall has been assessed and approved.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000659, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.4 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: ChatGPT extension on\nREQUIRED: external AI off unless approved\nDETAIL: Requests can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (ios):\n- Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted is non-compliant\", \"finding\": \"Apple Intelligence external integrations restricted is ChatGPT extension on instead of external AI off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Settings › Apple Intelligence & Siri › ChatGPT: turn off unless the organisation has approved it; the employer can restrict it with a configuration profile.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000531, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (ios):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI is pending\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001202, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain outbound [firewall.domain.outbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.outbound\", \"status\": \"Compliant\", \"title\": \"Domain outbound is compliant\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000916, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.domain.outbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.6 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI usage log retention [aiact.log_retention] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: depends on classification\nREQUIRED: >= 6 months for high-risk AI\nDETAIL: Required only if a tool is classified high-risk.\nREFERENCES:\n- ISO 27001:2022 A.8.15 Logging: Produce, protect and review logs of activities, exceptions and security events.\n- EU AI Act Art. 26(6) Deployers: keep automatically generated logs for at least six months: Deployers keep the logs a high-risk AI system generates, under their control, for at least six months unless other law says otherwise.\nGUIDANCE:\nwhy: Deployers of high-risk AI systems must keep the system's logs for at least six months.\nrisk: Medium - Missing logs prevent monitoring and investigation of high-risk AI use.\nfix (macos):\n- If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\nverify: Log retention is documented for each high-risk AI system.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.log_retention\", \"status\": \"Pending\", \"title\": \"AI usage log retention: needs a follow-up step\", \"finding\": \"Found: depends on classification. Required only if a tool is classified high-risk.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: Missing logs prevent monitoring and investigation of high-risk AI use.\"}, \"recommendation\": [\"If any listed tool is classified high-risk, keep its automatically generated logs under your control for at least six months.\"], \"references\": [\"ISO 27001:2022 A.8.15\", \"EU AI Act Art. 26(6)\"], \"verification\": \"Log retention is documented for each high-risk AI system.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000221, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.log_retention", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26200 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unauthorized employee local administrator accounts [admin.unauthorized] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: LAPTOP-6KE1NN4\\asmith, LAPTOP-6KE1NN4\\Administrator\nREQUIRED: none\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.18 Access rights: Access rights are granted, reviewed regularly and removed when no longer needed.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\nrisk: High - Malware running as an administrator can disable protection, install persistence and steal credentials.\nfix (windows):\n- Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\nverify: Re-run the audit; Unauthorized administrator accounts should read none."}, {"role": "assistant", "content": "{\"check_id\": \"admin.unauthorized\", \"status\": \"Non-Compliant\", \"title\": \"Unauthorized employee local administrator accounts is non-compliant\", \"finding\": \"The device reports LAPTOP-6KE1NN4\\\\asmith, LAPTOP-6KE1NN4\\\\Administrator, but the baseline requires none. Everyday user accounts with administrator rights let malware and mistakes change anything on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware running as an administrator can disable protection, install persistence and steal credentials.\"}, \"recommendation\": [\"Remove the listed accounts from the local Administrators group: Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>; give elevation through an approved process (e.g. Windows LAPS, an endpoint privilege management tool).\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.18\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Unauthorized administrator accounts should read none.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000732, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.unauthorized", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: On-access malware protection [malware.on_access] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: not running\nREQUIRED: on-access scanning\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: On-access scanning checks files when they are opened or written.\nrisk: Medium - Malicious files can be opened before any scheduled scan finds them.\nfix (linux):\n- Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\nverify: Re-run the audit; On-access malware protection should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.on_access\", \"status\": \"Non-Compliant\", \"title\": \"On-access malware protection: does not meet the baseline\", \"finding\": \"On-access malware protection is not running instead of on-access scanning. On-access scanning checks files when they are opened or written.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Malicious files can be opened before any scheduled scan finds them.\"}, \"recommendation\": [\"Configure OnAccessIncludePath in /etc/clamav/clamd.conf and enable clamonacc (systemctl enable --now clamav-clamonacc), or use an endpoint product with on-access scanning.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; On-access malware protection should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "malware.on_access", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Known vulnerabilities in iOS [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 15 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. iPadOS 26.7 checked against NVD iOS records. iPadOS 26.7 (15 CVEs, critical, actively exploited; fixed in 27.0; e.g. CVE-2026-20683, CVE-2026-43688). 32 medium/low.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (ios):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in iOS is non-compliant\", \"finding\": \"Found: 15 critical/high. Required: <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Settings › General › Software Update: install the newest iOS version; if the iPhone cannot run a version without the listed vulnerabilities, replace it.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (windows):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Non-Compliant\", \"title\": \"Data protection impact assessment is non-compliant\", \"finding\": \"The device reports no, but the baseline requires yes. High-risk processing needs a data protection impact assessment before it starts.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000159, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data protection impact assessment [org.dpia] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_DPIA in organisational.conf.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 35 Data protection impact assessment: Assess the impact before processing that is likely to result in a high risk to people.\nGUIDANCE:\nwhy: High-risk processing needs a data protection impact assessment before it starts.\nrisk: Medium - High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\nfix (linux):\n- List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\nverify: Answer ORG_DPIA=yes once DPIAs are done.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.dpia\", \"status\": \"Pending\", \"title\": \"Data protection impact assessment: decision pending\", \"finding\": \"Current state: not answered. Answer ORG_DPIA in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk processing without a DPIA breaches GDPR and leaves risks to people unassessed.\"}, \"recommendation\": [\"List processing that is likely high-risk (e.g. monitoring employees, large-scale sensitive data, new technologies such as AI) and complete a DPIA for each.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 35\"], \"verification\": \"Answer ORG_DPIA=yes once DPIAs are done.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000120, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.dpia", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Record of processing activities [org.ropa] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: partial\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\nGUIDANCE:\nwhy: The record of processing activities is the core GDPR accountability document.\nrisk: High - A missing or outdated record is a direct infringement and makes every other assessment harder.\nfix (windows):\n- Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\n- Assign an owner and review it at least yearly.\nverify: Answer ORG_ROPA=yes in organisational.conf once the record is current."}, {"role": "assistant", "content": "{\"check_id\": \"org.ropa\", \"status\": \"Non-Compliant\", \"title\": \"Record of processing activities: does not meet the baseline\", \"finding\": \"Record of processing activities is partial instead of yes. The record of processing activities is the core GDPR accountability document.\", \"risk\": {\"level\": \"High\", \"reason\": \"A missing or outdated record is a direct infringement and makes every other assessment harder.\"}, \"recommendation\": [\"Create or update the record of processing activities (purposes, categories of data and people, recipients, transfers, retention, security measures).\", \"Assign an owner and review it at least yearly.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 30\"], \"verification\": \"Answer ORG_ROPA=yes in organisational.conf once the record is current.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001410, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "org.ropa", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (windows):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved apps (Settings > Apps) or block them with AppLocker / Intune app control.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI) is compliant\", \"finding\": \"Found: none. This satisfies the baseline requirement of only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: NTP service yes, synchronised yes\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation is compliant\", \"finding\": \"Time synchronisation is set to NTP service yes, synchronised yes, in line with the required time synchronised.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000186, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: ai_classification\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAI TOOLS FOUND: Claude Code\nAPPROVED AI TOOLS: none\nPURPOSE OF USE: not stated\nTOOL FACTS:\n- Claude Code: AI coding assistant (command line); provider Anthropic (US); data: Code and prompts sent to the provider's cloud; default risk minimal; GDPR: Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\nRULES:\n- prohibited: Emotion recognition of employees or students (except for medical or safety reasons); Social scoring of people; Manipulative or exploitative techniques that cause significant harm; Untargeted scraping of facial images to build face recognition databases; Biometric categorisation inferring sensitive traits such as political views or sexual orientation\n- high-risk examples: Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions; Assessing creditworthiness or credit scores of individuals; Deciding access to education or grading students; Assessing eligibility for public benefits or essential services; Risk assessment and pricing in life and health insurance; Biometric identification of people\n- limited risk: Chatbots and AI-generated content: people must be told they are dealing with AI, and published deepfakes or AI-written public-interest text must be labelled (Art. 50).\n- minimal risk: Drafting, summarising, translation, coding help and similar general productivity use."}, {"role": "assistant", "content": "{\"tools\": [{\"tool\": \"Claude Code\", \"kind\": \"AI coding assistant (command line)\", \"approved\": false, \"default_risk_class\": \"minimal\", \"becomes_high_risk_if\": [\"Screening or ranking job applicants, or making promotion, task allocation or dismissal decisions\", \"Assessing creditworthiness or credit scores of individuals\"], \"prohibited_if\": [\"Emotion recognition of employees or students (except for medical or safety reasons)\"], \"gdpr\": \"Source code may contain personal data or secrets; use a commercial plan with a data processing agreement and keep secrets out of repositories.\", \"action\": \"Review: not approved. Approve it after a data protection and AI Act review, or remove it.\"}], \"purpose_needed\": true, \"log_retention\": \"Required (at least six months) only if a tool is used for a high-risk purpose.\", \"summary\": \"1 AI tool(s) found; default risk class minimal for general office use. The class rises to high-risk if a tool is used for an Annex III purpose such as screening job applicants. Not approved: Claude Code.\"}"}], "meta": {"task": "ai_classification", "os": "linux", "audit": 7000039, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (macos):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Data processing agreements\", \"finding\": \"Found: no. Required: yes. Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\", \"risk\": {\"level\": \"High\", \"reason\": \"Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000582, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Microsoft Defender Antimalware [malware.service] in COMPLIANCE RESULTS\nSTATUS: NotApplicable\nFOUND: Defender in Passive Mode\nREQUIRED: running\nDETAIL: Another antivirus is primary: Sophos Endpoint\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The malware protection service must be running for protection to work.\nrisk: High - A stopped protection service leaves the device unprotected.\nfix (windows):\n- Re-enable Microsoft Defender: check that no Group Policy sets 'Turn off Microsoft Defender Antivirus', then restart the device.\nverify: Re-run the audit; the malware protection service should be Compliant.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"malware.service\", \"status\": \"NotApplicable\", \"title\": \"Microsoft Defender Antimalware does not apply here\", \"finding\": \"Not applicable on this device: Another antivirus is primary: Sophos Endpoint.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001002, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.service", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: Ubuntu 24.04.3 LTS 24.04 (linux)\nHOSTNAME: dev-lt-65\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=42, Non-Compliant=8, NotApplicable=2, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Non-Compliant | PASS_MAX_DAYS 99999\n- password.min_age | Minimum password age | Compliant | 1 days\n- password.complexity | Password complexity | Compliant | minclass=0, required classes=4\n- lockout.threshold | Account lockout threshold | Compliant | 4\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 15 minutes\n- malware.antivirus | Antivirus | Compliant | ClamAV (clamav-daemon)\n- malware.service | ClamAV malware protection service | Compliant | clamav-daemon\n- malware.signature_updates | Malware signature updates | Compliant | clamav-freshclam\n- malware.signatures | Malware signatures | Compliant | 3 days old\n- malware.on_access | On-access malware protection | Compliant | clamonacc running\n- firewall.enabled | Firewall | Compliant | ufw active\n- firewall.inbound | Firewall inbound policy | Compliant | deny\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.timeout | Screen lock timeout | Compliant | within limit\n- screen_lock.password | Screen lock password requirement | Compliant | required\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Non-Compliant | allow\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | rschmit\n- user.account_type | Signed-in user account type | Compliant | Standard (ewagner)\n- encryption.enabled | Encryption | Compliant | root filesystem on LUKS (dm_crypt-0)\n- encryption.protection | Disk encryption protection | Compliant | LUKS2 aes-xts-plain64 512-bit key\n- encryption.recovery | Disk encryption recovery | Compliant | 2 key slots, 0 unlock tokens\n- vuln.pending_updates | Pending security updates | Non-Compliant | 36 pending\n- vuln.outdated_apps | Outdated third-party applications | Compliant | 0 outdated\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Compliant | 0 critical/high\n- gdpr.auto_updates | Automatic security updates | Compliant | unattended-upgrades installed=1, enabled=1\n- gdpr.updates_recent | Security updates installed recently | Compliant | 6 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | Ubuntu 24.04.3 LTS\n- gdpr.security_logging | Security event logging | Compliant | auditd running, journal persistent\n- gdpr.time_sync | Time synchronisation | Compliant | NTP service yes, synchronised yes\n- gdpr.backup_configured | Backup configured | Compliant | restic-backup.timer\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Non-Compliant | USB storage writable without enforced encryption\n- gdpr.guest_account | Guest account disabled | Compliant | no guest login\n- gdpr.remote_access | Remote access services restricted | Non-Compliant | chromeremotedesktop, smb\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | whoopsie error reporting on\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | GPT4All, Ollama, Alpaca (Ollama client)\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | GPT4All, Ollama, Alpaca (Ollama client)\n- aiact.risk_classification | AI system risk classification | Pending | GPT4All, Ollama, Alpaca (Ollama client)\n- aiact.os_ai_features | Operating system AI features restricted | NotApplicable | none built in\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Pending | not answered\n- org.breach_procedure | Breach response procedure | Compliant | yes\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Compliant | yes\n- org.human_oversight | Human oversight for high-risk AI | Compliant | yes\n- org.ai_transparency | AI transparency to users | Compliant | yes\nRISK LEVELS (from guidance):\n- password.expiration: Low\n- usb.write: Medium\n- admin.unauthorized: High\n- vuln.pending_updates: High\n- gdpr.removable_encryption: Medium\n- gdpr.remote_access: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\nSCORE (computed): compliant=42, assessed=50, percent=84\nTOP GAPS (computed, highest risk first):\n- admin.unauthorized | Unauthorized employee administrator accounts | High | rschmit\n- vuln.pending_updates | Pending security updates | High | 36 pending\n- gdpr.remote_access | Remote access services restricted | High | chromeremotedesktop, smb\n- usb.write | USB storage write | Medium | allow\n- gdpr.removable_encryption | Removable media encryption | Medium | USB storage writable without enforced encryption\nFIX GUIDANCE (highest-risk gaps):\n- admin.unauthorized: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\n- vuln.pending_updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\n- gdpr.remote_access: Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools."}, {"role": "assistant", "content": "{\"device\": \"dev-lt-65 - Ubuntu 24.04.3 LTS 24.04\", \"score\": {\"compliant\": 42, \"assessed\": 50, \"percent\": 84}, \"counts\": {\"Compliant\": 42, \"Non-Compliant\": 8, \"NotApplicable\": 2, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"rschmit\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"36 pending\"}, {\"check_id\": \"gdpr.remote_access\", \"name\": \"Remote access services restricted\", \"risk\": \"High\", \"found\": \"chromeremotedesktop, smb\"}, {\"check_id\": \"usb.write\", \"name\": \"USB storage write\", \"risk\": \"Medium\", \"found\": \"allow\"}, {\"check_id\": \"gdpr.removable_encryption\", \"name\": \"Removable media encryption\", \"risk\": \"Medium\", \"found\": \"USB storage writable without enforced encryption\"}], \"summary\": \"dev-lt-65 meets 42 of 50 assessed checks (84%). The most serious gaps are Unauthorized employee administrator accounts, Pending security updates and Remote access services restricted. 4 items are pending.\", \"next_steps\": [\"Unauthorized employee administrator accounts: Remove the listed accounts from the sudo/wheel groups (gpasswd -d <account> sudo) and remove their entries from /etc/sudoers.d.\", \"Pending security updates: Install pending security updates: apt update && apt upgrade (RHEL: dnf upgrade --security), then reboot if a new kernel was installed.\", \"Remote access services restricted: Disable unapproved services, e.g. systemctl disable --now ssh xrdp smbd, and remove unapproved remote access tools.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "linux", "audit": 7000154, "profile": "hardened", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] USB storage write on Linux: USB storage write on Linux. Why it matters: Writing to USB drives is a common way for data to leave the organisation, often unencrypted. Risk if missing: Medium - Personal or confidential data can be copied to an unmanaged, unencrypted drive and lost. Requirement: Effective USB storage write access equals USB_WRITE (allow/deny). How to fix on Linux: (1) Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule. How to verify: Re-run the audit; USB storage write should read deny. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.12 (Data leakage prevention); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[2] USB storage read on Linux: USB storage read on Linux. Why it matters: The baseline sets whether USB storage may be read; reading lets malware and unapproved data enter. Risk if missing: Low - The setting differs from the organisation's agreed USB policy. Requirement: Effective USB storage read access equals USB_READ (allow/deny). How to fix on Linux: (1) To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or use USBGuard. How to verify: Re-run the audit; USB storage read should read allow. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.12 (Data leakage prevention); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[3] USB storage device access on Linux: USB storage device access on Linux. Why it matters: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives. Risk if missing: Low - The device access setting differs from the organisation's agreed USB policy. Requirement: Whether USB mass storage devices can connect at all equals USB_DEVICE_ACCESS (allow/deny): driver, device-installation policy, USBGuard or MDM restriction. How to fix on Linux: (1) To block USB storage completely: echo 'install usb-storage /bin/false' > /etc/modprobe.d/usb-storage.conf, or enable USBGuard with ImplicitPolicyTarget=block. How to verify: Re-run the audit; USB storage device access should read allow. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.12 (Data leakage prevention); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n[4] Removable media encryption on Linux: Removable media encryption on Linux. Why it matters: Personal data copied to removable media must be encrypted, or copying must be blocked. Risk if missing: Medium - A lost unencrypted USB drive with personal data is a reportable breach. Requirement: Data cannot be written to unencrypted removable drives. How to fix on Linux: (1) Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives. How to verify: Re-run the audit; Removable media encryption should be Compliant. Related controls: ISO 27001:2022 A.7.10 (Storage media); ISO 27001:2022 A.8.24 (Use of cryptography); GDPR Art. 32(1)(a) (Security of processing: pseudonymisation and encryption); GDPR Art. 34 (Communication of a personal data breach to the data subject).\n\nMY AUDIT: Ubuntu 24.04.3 LTS 24.04 (linux), audited 2026-10-03; 32 of 46 assessed checks compliant; 14 gaps, 7 pending, 1 not read. Gaps: Password history: 10; Password expiration: PASS_MAX_DAYS 90; accounts: jdoe (99999); Account lockout duration: 10 minutes; Antivirus: none running; Malware signatures: no signature database; On-access malware protection: not running; Screen lock timeout: too long for: asmith (3600s); USB storage write: allow.\n\nQUESTION: How do I fix usb storage write?"}, {"role": "assistant", "content": "Your audit found USB storage write: allow. Writing to USB drives is a common way for data to leave the organisation, often unencrypted. To fix it [1]:\n1. Make USB storage read-only with a udev rule, or block it with USBGuard or the usb-storage modprobe rule.\n\nThen run the audit again to confirm."}], "meta": {"task": "chat", "os": "linux", "status": "audit", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 18.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: NotApplicable\nFOUND: no AI tools found\nREQUIRED: every AI tool classified\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (ios):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This check does not apply to this device or configuration."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"NotApplicable\", \"title\": \"AI system risk classification: not assessed on this device\", \"finding\": \"AI system risk classification is not assessed here: no AI tools found. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"Not applicable to this device.\"}, \"recommendation\": [], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000792, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "NotApplicable"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Google Play Protect [malware.antivirus] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Play Protect scanning off\nREQUIRED: on\nDETAIL: Installed apps are not checked for harmful behaviour.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Anti-malware software detects and blocks known malicious files and behaviour.\nrisk: High - Without working anti-malware, ransomware and info-stealers can run unchallenged.\nfix (android):\n- Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\nverify: Re-run the audit; Antivirus should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"malware.antivirus\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Google Play Protect\", \"finding\": \"The user reports Play Protect scanning off, but the baseline requires on. Anti-malware software detects and blocks known malicious files and behaviour. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Without working anti-malware, ransomware and info-stealers can run unchallenged.\"}, \"recommendation\": [\"Play Store › profile picture › Play Protect › Settings: turn on 'Scan apps with Play Protect'.\"], \"references\": [\"ISO 27001:2022 A.8.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Antivirus should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000873, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "malware.antivirus", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Usage & diagnostics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: usage & diagnostics off\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (android):\n- Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Pending\", \"title\": \"Usage & diagnostics sharing minimised: decision pending\", \"finding\": \"Found: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Google › All services › Usage & diagnostics: turn off; also turn off the phone maker's diagnostics sharing.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000521, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Domain inbound [firewall.domain.inbound] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: block\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultInboundAction Block.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.domain.inbound\", \"status\": \"Compliant\", \"title\": \"Domain inbound is compliant\", \"finding\": \"Found: block. This satisfies the baseline requirement of block.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000052, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "firewall.domain.inbound", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: summary\nDEVICE: macOS 14.7.8 (macos)\nHOSTNAME: MKT-MacBook-08\nAUDIT RUN AS ADMIN: yes\nCOUNTS: Compliant=34, Non-Compliant=14, NotApplicable=4, Pending=4, Error=0\nRESULTS (id | name | status | found):\n- password.min_length | Minimum password length | Compliant | 16\n- password.history | Password history | Compliant | 24\n- password.expiration | Password expiration | Compliant | 365 days\n- password.min_age | Minimum password age | NotApplicable | no such setting\n- password.complexity | Password complexity | Non-Compliant | not enforced\n- lockout.threshold | Account lockout threshold | Compliant | 3\n- lockout.duration | Account lockout duration | Compliant | until an administrator unlocks\n- lockout.observation | Lockout observation period | Compliant | 30 minutes\n- malware.antivirus | Antivirus | Compliant | XProtect 5253\n- malware.service | Apple malware protection service | Compliant | XProtect scanning on; Gatekeeper on\n- malware.signatures | Malware protection signatures | Compliant | XProtect 5343, 24 days old\n- malware.signature_updates | Malware security updates | Compliant | security responses and system data files install automatically\n- malware.on_access | On-access malware protection | NotApplicable | XProtect only\n- firewall.enabled | Firewall | Non-Compliant | off\n- firewall.inbound | Firewall inbound policy | Non-Compliant | firewall off, stealth mode off\n- firewall.outbound | Firewall outbound policy | Compliant | allow\n- screen_lock.timeout | Screen lock timeout | Compliant | 125 seconds\n- screen_lock.enabled | Screen lock | Compliant | on\n- screen_lock.password | Screen lock password requirement | Non-Compliant | password after 900 seconds\n- usb.device_access | USB storage device access | Compliant | allow\n- usb.read | USB storage read | Compliant | allow\n- usb.write | USB storage write | Compliant | deny\n- admin.approved | Approved IT administrator | Compliant | itadmin\n- admin.unauthorized | Unauthorized employee administrator accounts | Non-Compliant | mkhan, jdoe\n- user.account_type | Signed-in user account type | Compliant | Standard (asmith)\n- encryption.enabled | Encryption | Compliant | FileVault on\n- encryption.protection | Disk encryption protection | Compliant | fully encrypted\n- encryption.recovery | Disk encryption recovery | Compliant | recovery key present (personal: true, institutional: false)\n- vuln.pending_updates | Pending security updates | Non-Compliant | 1 pending\n- vuln.outdated_apps | Outdated third-party applications | NotApplicable | not checked\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | Non-Compliant | 18 critical/high\n- gdpr.auto_updates | Automatic security updates | Non-Compliant | check=1 download=1 security=1 macOS-install=0\n- gdpr.updates_recent | Security updates installed recently | Compliant | 15 days ago\n- gdpr.os_supported | Supported operating system version | Compliant | macOS 14.7.8\n- gdpr.security_logging | Security event logging | Compliant | unified logging active\n- gdpr.time_sync | Time synchronisation | Compliant | network time on\n- gdpr.backup_configured | Backup configured | Compliant | bzserv\n- gdpr.backup_recent | Recent backup | Pending | date not readable\n- gdpr.removable_encryption | Removable media encryption | Compliant | writing to external drives blocked\n- gdpr.guest_account | Guest account disabled | Compliant | guest login off\n- gdpr.remote_access | Remote access services restricted | Compliant | none enabled\n- gdpr.telemetry | Diagnostic data minimised | Non-Compliant | share Mac analytics=1, share with app developers=0\n- gdpr.personal_data_discovery | Personal data discovery | NotApplicable | not run\n- aiact.inventory | AI software inventory | Compliant | Ollama, Claude\n- aiact.unapproved | Unapproved AI tools (shadow AI) | Non-Compliant | Ollama, Claude\n- aiact.risk_classification | AI system risk classification | Pending | Ollama, Claude\n- aiact.os_ai_features | Apple Intelligence external integrations restricted | Compliant | restricted by profile\n- aiact.log_retention | AI usage log retention | Pending | depends on classification\n- org.ropa | Record of processing activities | Compliant | yes\n- org.dpia | Data protection impact assessment | Non-Compliant | partial\n- org.breach_procedure | Breach response procedure | Non-Compliant | no\n- org.processor_agreements | Data processing agreements | Compliant | yes\n- org.ai_register | AI register | Compliant | yes\n- org.ai_literacy | AI literacy training | Non-Compliant | no\n- org.human_oversight | Human oversight for high-risk AI | Non-Compliant | no\n- org.ai_transparency | AI transparency to users | Pending | not answered\nRISK LEVELS (from guidance):\n- password.complexity: Medium\n- firewall.enabled: High\n- firewall.inbound: High\n- screen_lock.password: Medium\n- admin.unauthorized: High\n- vuln.pending_updates: High\n- vuln.known_vulnerabilities: High\n- gdpr.auto_updates: High\n- gdpr.telemetry: Low\n- aiact.unapproved: Medium\n- org.dpia: Medium\n- org.breach_procedure: High\n- org.ai_literacy: Medium\n- org.human_oversight: Medium\nSCORE (computed): compliant=34, assessed=48, percent=71\nTOP GAPS (computed, highest risk first):\n- firewall.enabled | Firewall | High | off\n- firewall.inbound | Firewall inbound policy | High | firewall off, stealth mode off\n- admin.unauthorized | Unauthorized employee administrator accounts | High | mkhan, jdoe\n- vuln.pending_updates | Pending security updates | High | 1 pending\n- vuln.known_vulnerabilities | Known vulnerabilities in installed software | High | 18 critical/high\nFIX GUIDANCE (highest-risk gaps):\n- firewall.enabled: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\n- firewall.inbound: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\n- admin.unauthorized: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin."}, {"role": "assistant", "content": "{\"device\": \"MKT-MacBook-08 - macOS 14.7.8\", \"score\": {\"compliant\": 34, \"assessed\": 48, \"percent\": 71}, \"counts\": {\"Compliant\": 34, \"Non-Compliant\": 14, \"NotApplicable\": 4, \"Pending\": 4, \"Error\": 0}, \"top_gaps\": [{\"check_id\": \"firewall.enabled\", \"name\": \"Firewall\", \"risk\": \"High\", \"found\": \"off\"}, {\"check_id\": \"firewall.inbound\", \"name\": \"Firewall inbound policy\", \"risk\": \"High\", \"found\": \"firewall off, stealth mode off\"}, {\"check_id\": \"admin.unauthorized\", \"name\": \"Unauthorized employee administrator accounts\", \"risk\": \"High\", \"found\": \"mkhan, jdoe\"}, {\"check_id\": \"vuln.pending_updates\", \"name\": \"Pending security updates\", \"risk\": \"High\", \"found\": \"1 pending\"}, {\"check_id\": \"vuln.known_vulnerabilities\", \"name\": \"Known vulnerabilities in installed software\", \"risk\": \"High\", \"found\": \"18 critical/high\"}], \"summary\": \"MKT-MacBook-08 meets 34 of 48 assessed checks (71%). The most serious gaps are Firewall, Firewall inbound policy and Unauthorized employee administrator accounts. 4 items are pending.\", \"next_steps\": [\"Firewall: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on, or enforce the firewall by configuration profile.\", \"Firewall inbound policy: Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\", \"Unauthorized employee administrator accounts: Demote the listed accounts: sudo dseditgroup -o edit -d <account> -t user admin.\", \"Answer the 1 open organisational questions in organisational.conf.\", \"Classify the AI tools found on this device and record them in the AI register.\"]}"}], "meta": {"task": "summary", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: App updates [vuln.outdated_apps] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: automatic app updates off\nREQUIRED: automatic app updates on\nDETAIL: Apps keep known security bugs until they are updated by hand.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS.\nrisk: Medium - Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\nfix (android):\n- Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\nverify: Re-run the audit; outdated applications should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.outdated_apps\", \"status\": \"Non-Compliant\", \"title\": \"Gap: App updates\", \"finding\": \"The user reports automatic app updates off, but the baseline requires automatic app updates on. Browsers, office suites, PDF readers and meeting apps are common attack targets and need updates like the OS. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Outdated applications contain known vulnerabilities that attackers exploit through documents and web pages.\"}, \"recommendation\": [\"Play Store › profile picture › Settings › Network preferences › Auto-update apps: Over any network or Over Wi-Fi only; then Manage apps & device › Update all.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; outdated applications should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000972, "profile": "typical", "dataset_version": "0.5.0", "check_id": "vuln.outdated_apps", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 18.7.9 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Notification previews hidden when locked [gdpr.lock_screen_previews] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: Always\nREQUIRED: When Unlocked or Never\nDETAIL: Message and email content can be read on the locked screen.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 5(1)(f) Integrity and confidentiality: Personal data must be protected against unauthorised access, loss and damage with appropriate measures.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Previews show message and email content, often personal data, to anyone who sees the locked phone.\nrisk: Low - Personal data in notifications can be read without unlocking the phone.\nfix (ios):\n- Settings › Notifications › Show Previews: choose When Unlocked or Never.\nverify: Re-run the phone check; previews should be hidden when the phone is locked."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.lock_screen_previews\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Notification previews hidden when locked\", \"finding\": \"Found: Always. Required: When Unlocked or Never. Previews show message and email content, often personal data, to anyone who sees the locked phone. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Personal data in notifications can be read without unlocking the phone.\"}, \"recommendation\": [\"Settings › Notifications › Show Previews: choose When Unlocked or Never.\"], \"references\": [\"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 5(1)(f)\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; previews should be hidden when the phone is locked.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000660, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.lock_screen_previews", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Backup configured [gdpr.backup_configured] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: none\nREQUIRED: backup configured\nREFERENCES:\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: Backups make it possible to restore personal data after loss, theft, failure or ransomware.\nrisk: High - Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\nfix (linux):\n- Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\nverify: Re-run the audit; Backup configured should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.backup_configured\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Backup configured\", \"finding\": \"The device reports none, but the baseline requires backup configured. Backups make it possible to restore personal data after loss, theft, failure or ransomware.\", \"risk\": {\"level\": \"High\", \"reason\": \"Data that exists only on this device is lost if the device fails or is encrypted by ransomware.\"}, \"recommendation\": [\"Set up scheduled backups of /home with Deja Dup, restic or borg (or Timeshift for system snapshots plus a data backup).\"], \"references\": [\"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Backup configured should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001495, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.backup_configured", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Diagnostic data minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: popularity-contest on\nREQUIRED: usage reporting off\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (linux):\n- Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\nverify: Re-run the audit; Diagnostic data minimised should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Non-Compliant\", \"title\": \"Diagnostic data minimised is non-compliant\", \"finding\": \"Diagnostic data minimised is popularity-contest on instead of usage reporting off. Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\", \"risk\": {\"level\": \"Low\", \"reason\": \"More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Opt out of popularity-contest (dpkg-reconfigure popularity-contest) and disable whoopsie if present (systemctl disable --now whoopsie).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001287, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Signed-in user account type [user.account_type] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: Administrator (CONTOSO\\lmuller)\nREQUIRED: Standard\nDETAIL: The signed-in user is a member of the administrators group.\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Day-to-day work under an administrator account lets any malware or mistake change the whole system.\nrisk: High - Malware started by the signed-in user runs with full administrator rights.\nfix (windows):\n- Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\nverify: Re-run the audit; Signed-in user account type should read Standard."}, {"role": "assistant", "content": "{\"check_id\": \"user.account_type\", \"status\": \"Non-Compliant\", \"title\": \"Signed-in user account type is non-compliant\", \"finding\": \"The device reports Administrator (CONTOSO\\\\lmuller), but the baseline requires Standard. Day-to-day work under an administrator account lets any malware or mistake change the whole system.\", \"risk\": {\"level\": \"High\", \"reason\": \"Malware started by the signed-in user runs with full administrator rights.\"}, \"recommendation\": [\"Make the signed-in user a Standard user (Settings > Accounts > Other users > Change account type, or Remove-LocalGroupMember -SID S-1-5-32-544 -Member <account>) and use a separate approved admin account or LAPS for elevation.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Signed-in user account type should read Standard.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001210, "profile": "typical", "dataset_version": "0.5.0", "check_id": "user.account_type", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 14.7.8 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Firewall inbound policy [firewall.inbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: firewall off, stealth mode off\nREQUIRED: block\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\nrisk: High - Open inbound access lets attackers reach vulnerable services directly.\nfix (macos):\n- Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\nverify: Re-run the audit; the inbound policy should be block/deny."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.inbound\", \"status\": \"Non-Compliant\", \"title\": \"Firewall inbound policy: does not meet the baseline\", \"finding\": \"The device reports firewall off, stealth mode off, but the baseline requires block. Unsolicited inbound connections should be blocked unless a service is explicitly allowed.\", \"risk\": {\"level\": \"High\", \"reason\": \"Open inbound access lets attackers reach vulnerable services directly.\"}, \"recommendation\": [\"Turn on stealth mode: sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the inbound policy should be block/deny.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000700, "profile": "typical", "dataset_version": "0.5.0", "check_id": "firewall.inbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Data processing agreements [org.processor_agreements] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answered once by the organisation (organisational.conf); not part of the phone self-check.\nREFERENCES:\n- ISO 27001:2022 A.5.20 Addressing information security within supplier agreements: Agreements with suppliers set out the security requirements they must meet.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\nGUIDANCE:\nwhy: Every processor, including cloud and AI providers, needs a contract with the GDPR Article 28 terms.\nrisk: High - Sharing personal data with a processor without an agreement breaches GDPR.\nfix (android):\n- List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\nverify: Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.processor_agreements\", \"status\": \"Pending\", \"title\": \"Data processing agreements: needs a follow-up step\", \"finding\": \"Answered once by the organisation (organisational.conf); not part of the phone self-check. Current state: not answered.\", \"risk\": {\"level\": \"High\", \"reason\": \"Until resolved: Sharing personal data with a processor without an agreement breaches GDPR.\"}, \"recommendation\": [\"List all processors (including AI services used by staff) and sign a data processing agreement with each; check the transfer basis for non-EEA providers.\"], \"references\": [\"ISO 27001:2022 A.5.20\", \"GDPR Art. 28\"], \"verification\": \"Answer ORG_PROCESSOR_AGREEMENTS=yes once all agreements are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000060, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.processor_agreements", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 15.7.2 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Disk encryption recovery [encryption.recovery] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no FileVault\nREQUIRED: recovery key present\nREFERENCES:\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- ISO 27001:2022 A.8.13 Information backup: Keep backup copies of information and software and test that they can be restored.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 32(1)(c) Security of processing: restore availability and access after an incident: Be able to restore access to personal data in time after a physical or technical incident.\nGUIDANCE:\nwhy: A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\nrisk: Medium - Without a recovery key, an encrypted disk can become permanently unreadable.\nfix (macos):\n- Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\nverify: Re-run the audit; Disk encryption recovery should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"encryption.recovery\", \"status\": \"Non-Compliant\", \"title\": \"Disk encryption recovery is non-compliant\", \"finding\": \"Disk encryption recovery is no FileVault instead of recovery key present. A recovery key lets IT unlock the disk if the user forgets the password or the hardware changes.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a recovery key, an encrypted disk can become permanently unreadable.\"}, \"recommendation\": [\"Create or rotate a personal recovery key (sudo fdesetup changerecovery -personal) and escrow it to your MDM.\"], \"references\": [\"ISO 27001:2022 A.8.24\", \"ISO 27001:2022 A.8.13\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 32(1)(c)\"], \"verification\": \"Re-run the audit; Disk encryption recovery should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001139, "profile": "typical", "dataset_version": "0.5.0", "check_id": "encryption.recovery", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI transparency to users [org.ai_transparency] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- EU AI Act Art. 50 Transparency obligations for providers and deployers of certain AI systems: People must be told when they interact with an AI system, and deepfakes and certain AI-generated text must be disclosed.\nGUIDANCE:\nwhy: People must know when they deal with AI or AI-generated content.\nrisk: Low - Undisclosed AI use damages trust and breaches the AI Act transparency rules.\nfix (android):\n- Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\nverify: Answer ORG_AI_TRANSPARENCY=yes once the rules are in place."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_transparency\", \"status\": \"Non-Compliant\", \"title\": \"AI transparency to users: does not meet the baseline\", \"finding\": \"AI transparency to users is no instead of yes. People must know when they deal with AI or AI-generated content.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Undisclosed AI use damages trust and breaches the AI Act transparency rules.\"}, \"recommendation\": [\"Set rules for labelling AI chatbots and AI-generated or manipulated content, and include them in the AI acceptable-use policy.\"], \"references\": [\"EU AI Act Art. 50\"], \"verification\": \"Answer ORG_AI_TRANSPARENCY=yes once the rules are in place.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_transparency", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI literacy training [org.ai_literacy] in ORGANISATIONAL CONTROLS\nSTATUS: Compliant\nFOUND: yes\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.6.3 Information security awareness, education and training: People receive awareness and training suited to their role, and it is kept current.\n- EU AI Act Art. 4 AI literacy: Providers and deployers take measures so their staff using AI have a sufficient level of AI literacy.\nGUIDANCE:\nwhy: Since 2 February 2025, organisations must ensure staff using AI have sufficient AI literacy.\nrisk: Medium - Untrained staff misuse AI tools and share data they should not.\nfix (android):\n- Provide AI literacy training matched to how staff use AI (capabilities, limits, data protection, the AI acceptable-use policy) and keep records.\nverify: Answer ORG_AI_LITERACY=yes once training is delivered and recorded."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_literacy\", \"status\": \"Compliant\", \"title\": \"AI literacy training matches the required setting\", \"finding\": \"Found: yes. This satisfies the baseline requirement of yes.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.6.3\", \"EU AI Act Art. 4\"]}"}], "meta": {"task": "finding", "os": "android", "audit": 7000320, "profile": "typical", "dataset_version": "0.5.0", "check_id": "org.ai_literacy", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password expiration [password.expiration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: PASS_MAX_DAYS 90; accounts: jdoe (99999)\nREQUIRED: 1-365 days\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\nrisk: Low - A leaked password can stay valid indefinitely.\nfix (linux):\n- Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\n- Apply it to existing accounts: chage --maxdays 365 <user>\nverify: Re-run the audit; Password expiration should be between 1 and 365 days.\nnote: Current NIST guidance (SP 800-63B) advises against forced periodic changes when passwords are long and checked against breach lists. Set PASSWORD_MAX_AGE_DAYS=0 in baseline.conf if your policy follows that approach."}, {"role": "assistant", "content": "{\"check_id\": \"password.expiration\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Password expiration\", \"finding\": \"Found: PASS_MAX_DAYS 90; accounts: jdoe (99999). Required: 1-365 days. The baseline requires passwords to be changed within a set period, limiting how long a stolen password stays useful.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A leaked password can stay valid indefinitely.\"}, \"recommendation\": [\"Set PASS_MAX_DAYS 365 in /etc/login.defs (applies to new accounts).\", \"Apply it to existing accounts: chage --maxdays 365 <user>\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password expiration should be between 1 and 365 days.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7001243, "profile": "typical", "dataset_version": "0.5.0", "check_id": "password.expiration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: AI register [org.ai_register] in ORGANISATIONAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: no\nREQUIRED: yes\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: A register of AI systems is the basis for AI Act classification and deployer duties.\nrisk: Medium - Without a register, high-risk or prohibited AI use can go unnoticed.\nfix (macos):\n- Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\nverify: Answer ORG_AI_REGISTER=yes once the register exists."}, {"role": "assistant", "content": "{\"check_id\": \"org.ai_register\", \"status\": \"Non-Compliant\", \"title\": \"AI register: does not meet the baseline\", \"finding\": \"The device reports no, but the baseline requires yes. A register of AI systems is the basis for AI Act classification and deployer duties.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Without a register, high-risk or prohibited AI use can go unnoticed.\"}, \"recommendation\": [\"Create an AI register listing each AI system, owner, purpose, risk class and the data it uses; use the AI software inventory from this audit as input.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_AI_REGISTER=yes once the register exists.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.ai_register", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 12 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI software inventory [aiact.inventory] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: no AI tools found\nREQUIRED: inventory taken\nDETAIL: AI apps listed by the user.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- GDPR Art. 30 Records of processing activities: Keep a written record of the processing activities you carry out.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Deployers must know which AI systems they use before they can classify and govern them.\nrisk: Low - The inventory itself is information; the risk lies in unapproved or unclassified tools.\nfix (android):\n- Add the AI tools found on this device to the organisation's AI register, with owner and purpose.\nverify: The AI register lists every tool in the inventory."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.inventory\", \"status\": \"Compliant\", \"title\": \"AI software inventory: meets the baseline\", \"finding\": \"The user reports no AI tools found, which meets the requirement (inventory taken). Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"GDPR Art. 30\", \"EU AI Act Art. 26\"], \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000178, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.inventory", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 8.10 (Ootpa) 8.10 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: AI system risk classification [aiact.risk_classification] in EU AI ACT CONTROLS\nSTATUS: Pending\nFOUND: LM Studio, Alpaca (Ollama client), Jan\nREQUIRED: every AI tool classified\nDETAIL: Classified by the auditor model (Caveat classifier) in a later step.\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- EU AI Act Art. 5 Prohibited AI practices: Certain AI uses are banned, such as emotion recognition at work and in education (with narrow exceptions) and social scoring.\n- EU AI Act Art. 6 Classification rules for high-risk AI systems: An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III.\n- EU AI Act Annex III High-risk AI systems referred to in Article 6(2): Lists high-risk uses such as biometrics, critical infrastructure, education, employment and worker management, access to essential services, law enforcement, migration and justice.\nGUIDANCE:\nwhy: Each AI system must be checked against the banned practices and the high-risk uses in Annex III.\nrisk: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\nfix (linux):\n- Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\nverify: Every AI tool in the inventory has a classification in the AI register.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.risk_classification\", \"status\": \"Pending\", \"title\": \"AI system risk classification is pending\", \"finding\": \"Classified by the auditor model (Caveat classifier) in a later step. Current state: LM Studio, Alpaca (Ollama client), Jan.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards.\"}, \"recommendation\": [\"Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"EU AI Act Art. 5\", \"EU AI Act Art. 6\", \"EU AI Act Annex III\"], \"verification\": \"Every AI tool in the inventory has a classification in the AI register.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000100, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.risk_classification", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Debian GNU/Linux 12 (bookworm) 12 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Password complexity [password.complexity] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: pam_pwquality not enabled\nREQUIRED: 3 or more character classes\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Mixing character types makes passwords harder to guess.\nrisk: Medium - Simple passwords are cracked quickly by dictionary attacks.\nfix (linux):\n- Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\nverify: Re-run the audit; Password complexity should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"password.complexity\", \"status\": \"Non-Compliant\", \"title\": \"Password complexity: does not meet the baseline\", \"finding\": \"Password complexity is pam_pwquality not enabled instead of 3 or more character classes. Mixing character types makes passwords harder to guess.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Simple passwords are cracked quickly by dictionary attacks.\"}, \"recommendation\": [\"Enable pam_pwquality and set minclass = 3 (or 4) in /etc/security/pwquality.conf.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Password complexity should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000693, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "password.complexity", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 10 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (linux):\n- Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"The device reports 10 minutes, but the baseline requires >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set unlock_time = 900 (seconds) in /etc/security/faillock.conf.\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000955, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 27.0 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Apps only from the App Store [mobile.app_sources] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: apps from outside the App Store\nREQUIRED: App Store only\nDETAIL: Apps from outside the App Store skip Apple's review.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.8.7 Protection against malware: Protect against malware with technical controls supported by user awareness.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Official app stores review apps for malware; apps from other sources skip those checks.\nrisk: High - Side-loaded apps are a common way malware reaches phones.\nfix (ios):\n- Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\nverify: Re-run the phone check; apps should come only from the official store (or the employer)."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.app_sources\", \"status\": \"Non-Compliant\", \"title\": \"Apps only from the App Store: does not meet the baseline\", \"finding\": \"Found: apps from outside the App Store. Required: App Store only. Official app stores review apps for malware; apps from other sources skip those checks. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Side-loaded apps are a common way malware reaches phones.\"}, \"recommendation\": [\"Remove apps from alternative marketplaces or websites, and remove unknown profiles in Settings › General › VPN & Device Management; install work apps only from the App Store or the employer.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.8.7\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; apps should come only from the official store (or the employer).\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000588, "profile": "typical", "dataset_version": "0.5.0", "check_id": "mobile.app_sources", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Approved IT administrator [admin.approved] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: itadmin\nREQUIRED: itadmin\nREFERENCES:\n- ISO 27001:2022 A.8.2 Privileged access rights: Restrict and manage the allocation and use of administrator rights.\n- ISO 27001:2022 A.5.15 Access control: Grant access to information and systems based on business need, and document the rules.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A known, approved IT account should manage the device so that support does not depend on users' own admin rights.\nrisk: Medium - Without an approved admin account, IT tends to rely on employees' accounts having admin rights.\nfix (linux):\n- Add the approved IT account to the sudo (Debian/Ubuntu) or wheel (RHEL) group: usermod -aG sudo <account>.\nverify: Re-run the audit; Approved IT administrator should list the approved account."}, {"role": "assistant", "content": "{\"check_id\": \"admin.approved\", \"status\": \"Compliant\", \"title\": \"Approved IT administrator: meets the baseline\", \"finding\": \"The device reports itadmin, which meets the requirement (itadmin).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.2\", \"ISO 27001:2022 A.5.15\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000031, "profile": "typical", "dataset_version": "0.5.0", "check_id": "admin.approved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 10 Pro 10.0.19045 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public outbound [firewall.public.outbound] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: block\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The baseline fixes the default outbound policy so that it is consistent across devices.\nrisk: Low - A different outbound policy either breaks applications or deviates from the agreed design.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\nverify: Re-run the audit; the outbound policy should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.outbound\", \"status\": \"Non-Compliant\", \"title\": \"Public outbound: does not meet the baseline\", \"finding\": \"Public outbound is block instead of allow. The baseline fixes the default outbound policy so that it is consistent across devices.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A different outbound policy either breaks applications or deviates from the agreed design.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -DefaultOutboundAction Allow.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the outbound policy should read allow.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000023, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.outbound", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: USB storage device access [usb.device_access] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: allow\nREQUIRED: allow\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.12 Data leakage prevention: Apply measures that stop sensitive information leaving through unauthorised channels.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Blocking USB mass storage devices completely is the strongest protection against data leaving on, or malware arriving from, removable drives.\nrisk: Low - The device access setting differs from the organisation's agreed USB policy.\nfix (macos):\n- Block or allow external storage with your MDM (Restrictions / device control) or an endpoint DLP product.\nverify: Re-run the audit; USB storage device access should read allow."}, {"role": "assistant", "content": "{\"check_id\": \"usb.device_access\", \"status\": \"Compliant\", \"title\": \"USB storage device access is compliant\", \"finding\": \"Found: allow. This satisfies the baseline requirement of allow.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.12\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "usb.device_access", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.0.1 (macos)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock password requirement [screen_lock.password] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: password after off seconds\nREQUIRED: password within 5 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Unlocking the screen must require the user's password.\nrisk: Medium - A screen lock that needs no password does not protect anything.\nfix (macos):\n- System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\nverify: Re-run the audit; the password requirement should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.password\", \"status\": \"Non-Compliant\", \"title\": \"Screen lock password requirement is non-compliant\", \"finding\": \"Screen lock password requirement is password after off seconds instead of password within 5 seconds. Unlocking the screen must require the user's password.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A screen lock that needs no password does not protect anything.\"}, \"recommendation\": [\"System Settings > Lock Screen > 'Require password after screen saver begins or display is turned off' = Immediately (askForPasswordDelay = 0 by profile).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the password requirement should be Compliant.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7000332, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "screen_lock.password", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iOS 26.6 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: iOS security updates [vuln.pending_updates] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: iOS 26.6 (update to 26.7.1 available)\nREQUIRED: latest iOS security release\nDETAIL: Install iOS 26.7.1 or later in Settings › General › Software Update.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Security updates that are offered but not installed leave known, published vulnerabilities open.\nrisk: High - Attackers target published vulnerabilities quickly after the fix is released.\nfix (ios):\n- Settings › General › Software Update: install the latest iOS version offered.\nverify: Re-run the audit; pending security updates should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.pending_updates\", \"status\": \"Non-Compliant\", \"title\": \"Gap: iOS security updates\", \"finding\": \"Found: iOS 26.6 (update to 26.7.1 available). Required: latest iOS security release. Security updates that are offered but not installed leave known, published vulnerabilities open. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"High\", \"reason\": \"Attackers target published vulnerabilities quickly after the fix is released.\"}, \"recommendation\": [\"Settings › General › Software Update: install the latest iOS version offered.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; pending security updates should be 0 or fewer.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000836, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.pending_updates", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Red Hat Enterprise Linux 9.6 (Plow) 9.6 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Compliant\nFOUND: within limit\nREQUIRED: <= 300 seconds\nDETAIL: Users checked: ewagner\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (linux):\n- gsettings set org.gnome.desktop.session idle-delay 300 and gsettings set org.gnome.desktop.screensaver lock-delay 0; enforce with locked dconf settings.\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Compliant\", \"title\": \"Screen lock timeout is compliant\", \"finding\": \"Screen lock timeout is set to within limit, in line with the required <= 300 seconds.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000810, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22621 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Known vulnerabilities in installed software [vuln.known_vulnerabilities] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 62 critical/high\nREQUIRED: <= 0 critical/high\nDETAIL: Matched against offline OSV/NVD data from 2026-10-08. 15 applications inventoried, 11 recognised (31 common applications covered). Microsoft Edge 150.0.4078.80 (14 CVEs, critical; fixed in 153.0.4234.46; e.g. CVE-2026-57989, CVE-2026-65802); Zoom Workplace 6.4.13 (8 CVEs, critical; fixed in 7.0.5; e.g. CVE-2025-49460, CVE-2025-49461); VLC media player 3.0.19 (2 CVEs, critical; fixed in 3.0.20; e.g. CVE-2023-47359, CVE-2023-47360); Adobe Acrobat Reader (64-bit) 26.001.21411 (37 CVEs, high; fixed in 26.002.21901; e.g. CVE-2026-34622, CVE-2026-47911); Notepad++ (64-bit x64) 8.9.6.1 (1 CVE, high; e.g. CVE-2026-52884). 44 medium/low.\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\nrisk: High - Critical and high vulnerabilities often allow remote code execution or privilege escalation.\nfix (windows):\n- Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\n- Use winget upgrade or the vendor's installer to reach a fixed version.\nverify: Re-run the audit; critical or high vulnerabilities should be 0 or fewer."}, {"role": "assistant", "content": "{\"check_id\": \"vuln.known_vulnerabilities\", \"status\": \"Non-Compliant\", \"title\": \"Known vulnerabilities in installed software: does not meet the baseline\", \"finding\": \"The device reports 62 critical/high, but the baseline requires <= 0 critical/high. Installed software versions with published critical or high vulnerabilities give attackers a ready-made way in.\", \"risk\": {\"level\": \"High\", \"reason\": \"Critical and high vulnerabilities often allow remote code execution or privilege escalation.\"}, \"recommendation\": [\"Update or remove each listed software package to a version without the listed vulnerabilities; where no fix exists yet, apply the vendor's workaround or restrict the software's use.\", \"Use winget upgrade or the vendor's installer to reach a fixed version.\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; critical or high vulnerabilities should be 0 or fewer.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000258, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "vuln.known_vulnerabilities", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 26.04.1 LTS 26.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Time synchronisation [gdpr.time_sync] in GDPR TECHNICAL CONTROLS\nSTATUS: Compliant\nFOUND: NTP service yes, synchronised yes\nREQUIRED: time synchronised\nREFERENCES:\n- ISO 27001:2022 A.8.17 Clock synchronization: Synchronise system clocks to an approved time source.\n- GDPR Art. 33 Notification of a personal data breach to the supervisory authority: Report a personal data breach to the supervisory authority within 72 hours of becoming aware of it, unless it is unlikely to result in a risk.\nGUIDANCE:\nwhy: Accurate clocks are needed to put events in the right order during an investigation.\nrisk: Low - Wrong timestamps make incident timelines unreliable.\nfix (linux):\n- timedatectl set-ntp true (or enable chronyd).\nverify: Re-run the audit; Time synchronisation should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.time_sync\", \"status\": \"Compliant\", \"title\": \"Time synchronisation: meets the baseline\", \"finding\": \"The device reports NTP service yes, synchronised yes, which meets the requirement (time synchronised).\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.8.17\", \"GDPR Art. 33\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000815, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.time_sync", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.7 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Passcode strength [mobile.passcode_strength] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 4-digit code\nREQUIRED: >= 6 digits or alphanumeric\nDETAIL: A 4-digit code has only 10000 combinations.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.17 Authentication information: Passwords and other secrets are allocated, managed and protected under clear rules.\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost.\nrisk: Medium - A weak passcode lets someone who has the phone unlock it and read work data.\nfix (ios):\n- Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\nverify: Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric."}, {"role": "assistant", "content": "{\"check_id\": \"mobile.passcode_strength\", \"status\": \"Non-Compliant\", \"title\": \"Passcode strength is non-compliant\", \"finding\": \"The user reports 4-digit code, but the baseline requires >= 6 digits or alphanumeric. A short numeric code or a pattern can be guessed or seen over the user's shoulder; a longer code protects the data if the phone is lost. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A weak passcode lets someone who has the phone unlock it and read work data.\"}, \"recommendation\": [\"Settings › Face ID & Passcode › Change Passcode › Passcode Options: choose a 6-digit, custom numeric (7+ digits) or alphanumeric code.\"], \"references\": [\"ISO 27001:2022 A.5.17\", \"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the phone check; the passcode answer should be at least 6 digits or alphanumeric.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7001383, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "mobile.passcode_strength", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 22.04.5 LTS 22.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Unapproved AI tools (shadow AI) [aiact.unapproved] in EU AI ACT CONTROLS\nSTATUS: Compliant\nFOUND: none\nREQUIRED: only: none\nREFERENCES:\n- ISO 27001:2022 A.5.9 Inventory of information and other associated assets: Know which information assets, devices and software exist and who owns them, and keep that list current.\n- ISO 27001:2022 A.8.19 Installation of software on operational systems: Control which software users may install.\n- GDPR Art. 28 Processor: Use only processors that give sufficient guarantees, under a written contract with required terms.\n- GDPR Art. 44 General principle for transfers: Transfer personal data outside the EEA only under the conditions of Chapter V.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: AI tools used without approval may send work and personal data to providers without contracts or a transfer basis.\nrisk: Medium - Shadow AI can leak confidential or personal data and creates unassessed AI Act and GDPR obligations.\nfix (linux):\n- Decide for each listed tool: approve it (add to APPROVED_AI_TOOLS after a data protection and AI Act review, with a data processing agreement where needed) or remove it.\n- Publish an AI acceptable-use policy so staff know which tools are allowed.\n- Remove unapproved tools with the package manager and restrict installation rights.\nverify: Re-run the audit; Unapproved AI tools should read none."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.unapproved\", \"status\": \"Compliant\", \"title\": \"Unapproved AI tools (shadow AI): meets the baseline\", \"finding\": \"Unapproved AI tools (shadow AI) is set to none, in line with the required only: none.\", \"risk\": {\"level\": \"None\", \"reason\": \"The setting meets the baseline.\"}, \"recommendation\": [\"Keep the current setting and confirm it at the next audit.\"], \"references\": [\"ISO 27001:2022 A.5.9\", \"ISO 27001:2022 A.8.19\", \"GDPR Art. 28\", \"GDPR Art. 44\", \"EU AI Act Art. 26\"]}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000766, "profile": "typical", "dataset_version": "0.5.0", "check_id": "aiact.unapproved", "status": "Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: macOS 26.5.2 (macos)\nAUDIT RUN AS ADMIN: no\nCHECK: Apple Intelligence external integrations restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: not restricted\nREQUIRED: restricted by profile\nDETAIL: No profile sets allowExternalIntelligenceIntegrations to false.\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (macos):\n- Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"Apple Intelligence external integrations restricted is non-compliant\", \"finding\": \"Apple Intelligence external integrations restricted is not restricted instead of restricted by profile. Built-in OS AI features can capture screen content or send text to external AI services.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"Deploy a Restrictions profile with allowExternalIntelligenceIntegrations = false (and restrict other Apple Intelligence features your policy does not allow).\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\"}"}], "meta": {"task": "finding", "os": "macos", "audit": 7001313, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Enterprise 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Public firewall [firewall.public.enabled] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: off\nREQUIRED: on\nREFERENCES:\n- ISO 27001:2022 A.8.20 Networks security: Secure, manage and control networks and network devices to protect information.\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: A host firewall blocks unwanted network connections to the device, especially on public networks.\nrisk: High - Network services on the device are exposed to anyone on the same network.\nfix (windows):\n- Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\nverify: Re-run the audit; the firewall should be on."}, {"role": "assistant", "content": "{\"check_id\": \"firewall.public.enabled\", \"status\": \"Non-Compliant\", \"title\": \"Public firewall: does not meet the baseline\", \"finding\": \"The device reports off, but the baseline requires on. A host firewall blocks unwanted network connections to the device, especially on public networks.\", \"risk\": {\"level\": \"High\", \"reason\": \"Network services on the device are exposed to anyone on the same network.\"}, \"recommendation\": [\"Set-NetFirewallProfile -Profile Domain,Private,Public -Enabled True, or enforce it with Group Policy/Intune.\"], \"references\": [\"ISO 27001:2022 A.8.20\", \"ISO 27001:2022 A.8.1\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; the firewall should be on.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000293, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "firewall.public.enabled", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.26100 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Account lockout duration [lockout.duration] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: 5 minutes\nREQUIRED: >= 15 minutes\nREFERENCES:\n- ISO 27001:2022 A.8.5 Secure authentication: Authentication methods and procedures resist guessing, brute force and misuse.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The lockout must last long enough to slow down repeated guessing.\nrisk: Low - A short lockout lets attackers resume guessing almost at once.\nfix (windows):\n- Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\n- For a stand-alone device: net accounts /lockoutduration:15\nverify: Re-run the audit; Account lockout duration should be 15 minutes or more."}, {"role": "assistant", "content": "{\"check_id\": \"lockout.duration\", \"status\": \"Non-Compliant\", \"title\": \"Account lockout duration is non-compliant\", \"finding\": \"Account lockout duration is 5 minutes instead of >= 15 minutes. The lockout must last long enough to slow down repeated guessing.\", \"risk\": {\"level\": \"Low\", \"reason\": \"A short lockout lets attackers resume guessing almost at once.\"}, \"recommendation\": [\"Set Group Policy: Account Lockout Policy > Account lockout duration = 15 minutes or more.\", \"For a stand-alone device: net accounts /lockoutduration:15\"], \"references\": [\"ISO 27001:2022 A.8.5\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Account lockout duration should be 15 minutes or more.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7001430, "profile": "typical", "dataset_version": "0.5.0", "check_id": "lockout.duration", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: iPadOS 26.5 (ios)\nAUDIT RUN AS ADMIN: no\nCHECK: Analytics sharing minimised [gdpr.telemetry] in GDPR TECHNICAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: analytics sharing off\nDETAIL: The user could not find or confirm this setting.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\nGUIDANCE:\nwhy: Operating systems can send optional usage and diagnostic data, which may contain personal data, to the vendor.\nrisk: Low - More personal data than necessary leaves the organisation.\nfix (ios):\n- Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\nverify: Re-run the audit; Diagnostic data minimised should be Compliant.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.telemetry\", \"status\": \"Pending\", \"title\": \"Analytics sharing minimised is pending\", \"finding\": \"Current state: not answered. The user could not find or confirm this setting. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Low\", \"reason\": \"Until resolved: More personal data than necessary leaves the organisation.\"}, \"recommendation\": [\"Settings › Privacy & Security › Analytics & Improvements: turn off Share iPhone Analytics.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\"], \"verification\": \"Re-run the audit; Diagnostic data minimised should be Compliant.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "ios", "audit": 7000771, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "gdpr.telemetry", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Screen lock timeout [screen_lock.timeout] in COMPLIANCE RESULTS\nSTATUS: Non-Compliant\nFOUND: no machine inactivity limit; too long or off for: FABRIKAM\\asmith (600s, active=0)\nREQUIRED: <= 300 seconds\nREFERENCES:\n- ISO 27001:2022 A.8.1 User endpoint devices: Protect information stored on, processed by or reachable from user devices.\n- ISO 27001:2022 A.7.7 Clear desk and clear screen: Unattended screens and papers must not expose information.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\nGUIDANCE:\nwhy: The shorter the idle time before locking, the smaller the window for misuse.\nrisk: Medium - A long or missing timeout leaves unattended devices open.\nfix (windows):\n- Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\nverify: Re-run the audit; Screen lock timeout should be 300 seconds or less."}, {"role": "assistant", "content": "{\"check_id\": \"screen_lock.timeout\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Screen lock timeout\", \"finding\": \"Screen lock timeout is no machine inactivity limit; too long or off for: FABRIKAM\\\\asmith (600s, active=0) instead of <= 300 seconds. The shorter the idle time before locking, the smaller the window for misuse.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A long or missing timeout leaves unattended devices open.\"}, \"recommendation\": [\"Set Group Policy: Computer Configuration > Windows Settings > Security Settings > Local Policies > Security Options > Interactive logon: Machine inactivity limit = 300 seconds (or less).\"], \"references\": [\"ISO 27001:2022 A.8.1\", \"ISO 27001:2022 A.7.7\", \"GDPR Art. 32(1)(b)\"], \"verification\": \"Re-run the audit; Screen lock timeout should be 300 seconds or less.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000415, "profile": "typical", "dataset_version": "0.5.0", "check_id": "screen_lock.timeout", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Microsoft Windows 11 Pro 10.0.22631 (windows)\nAUDIT RUN AS ADMIN: yes\nCHECK: Security updates installed recently [gdpr.updates_recent] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: 235 days ago\nREQUIRED: <= 30 days ago\nDETAIL: Last update: 2026-09 Cumulative Update for Windows 11 (2026-02-10)\nREFERENCES:\n- ISO 27001:2022 A.8.8 Management of technical vulnerabilities: Find technical weaknesses in time, judge the exposure and fix them.\n- GDPR Art. 32(1)(b) Security of processing: confidentiality, integrity, availability and resilience: Ensure the ongoing confidentiality, integrity, availability and resilience of systems that process personal data.\n- GDPR Art. 32(1)(d) Security of processing: regular testing and evaluation: Regularly test, assess and evaluate whether security measures work.\nGUIDANCE:\nwhy: Recent installs show that patching is actually happening, not only configured.\nrisk: High - A device not updated for a long time is likely missing critical security fixes.\nfix (windows):\n- Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\nverify: Re-run the audit; the last update should be within 30 days."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.updates_recent\", \"status\": \"Non-Compliant\", \"title\": \"Gap: Security updates installed recently\", \"finding\": \"Security updates installed recently is 235 days ago instead of <= 30 days ago. Recent installs show that patching is actually happening, not only configured.\", \"risk\": {\"level\": \"High\", \"reason\": \"A device not updated for a long time is likely missing critical security fixes.\"}, \"recommendation\": [\"Run Windows Update now (Settings > Windows Update > Check for updates) and investigate why updates stopped (paused updates, broken update service, metered connection).\"], \"references\": [\"ISO 27001:2022 A.8.8\", \"GDPR Art. 32(1)(b)\", \"GDPR Art. 32(1)(d)\"], \"verification\": \"Re-run the audit; the last update should be within 30 days.\"}"}], "meta": {"task": "finding", "os": "windows", "audit": 7000981, "profile": "typical", "dataset_version": "0.5.0", "check_id": "gdpr.updates_recent", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis security assistant. You help people secure their computers and phones and understand ISO 27001, GDPR and the EU AI Act. Answer in plain language that a non-technical person can follow, using only the CONTEXT and MY AUDIT below, and cite the passages you use as [1], [2]. If the CONTEXT does not answer the question, say you do not know rather than guessing. If the question is about the user's own device and MY AUDIT is \"none\", explain that you cannot see their device and suggest running an audit. Never invent settings, versions or numbers."}, {"role": "user", "content": "CONTEXT:\n[1] AI system risk classification on macOS: AI system risk classification on macOS. Why it matters: Each AI system must be checked against the banned practices and the high-risk uses in Annex III. Risk if missing: Medium - An unclassified tool could be used for a high-risk or prohibited purpose without the required safeguards. Requirement: Each AI tool found is classified (prohibited / high-risk / limited / minimal). Done by the auditor model: Pending in the collector output, NotApplicable when no AI tool is found. How to fix on macOS: (1) Record the purpose for which each listed tool is used and classify it (prohibited, high-risk, limited-risk transparency, minimal). The auditor model does this from the inventory and the stated purpose. How to verify: Every AI tool in the inventory has a classification in the AI register. Related controls: ISO 27001:2022 A.5.9 (Inventory of information and other associated assets); EU AI Act Art. 5 (Prohibited AI practices); EU AI Act Art. 6 (Classification rules for high-risk AI systems); EU AI Act Annex III (High-risk AI systems referred to in Article 6(2)).\n[2] How many checks the audit runs: The audit catalog has 74 checks: Windows 64, Linux 56, macOS 56, iPhone 29, Android 30. They cover passwords and lockout, malware protection, firewall, screen lock, USB storage, administrator accounts, disk encryption, updates and known vulnerabilities, GDPR technical controls (backups, logging, remote access, telemetry and more), EU AI Act controls (AI tools found, shadow AI, risk classification) and 8 organisational controls.\n[3] EU AI Act Art. 6 Classification rules for high-risk AI systems: EU AI Act Art. 6 (Classification rules for high-risk AI systems): An AI system is high-risk if it is a safety component of regulated products or is used for a purpose listed in Annex III. An auditor looks for: A risk classification for each AI system in use. On endpoints: Classification depends on what the tool is used for, not only on which tool it is.\n[4] Lockout observation period on macOS: Lockout observation period on macOS. Why it matters: Failed attempts must be counted over a long enough window to detect slow guessing. Risk if missing: Low - A short window lets attackers pace their guesses to avoid lockout. Requirement: Failed-attempt counter kept for >= LOCKOUT_OBSERVATION_MIN_MINUTES minutes. How to fix on macOS: (1) On macOS the counter resets when the lockout ends; set minutesUntilFailedLoginReset = 15 in the Passcode profile. How to verify: Re-run the audit; Lockout observation period should be 15 minutes or more. Related controls: ISO 27001:2022 A.8.5 (Secure authentication); GDPR Art. 32(1)(b) (Security of processing: confidentiality, integrity, availability and resilience).\n\nMY AUDIT: none\n\nMY DEVICE: macOS 13.6 (macos) on a Mac (Intel), Safari 18.6; detected by the browser, security settings not checked.\n\nQUESTION: Which ISO 27001, GDPR or EU AI Act rules cover lockout observation period?"}, {"role": "assistant", "content": "Lockout observation period relates to ISO 27001:2022 A.8.5, GDPR Art. 32(1)(b) [4]. Meeting the setting is evidence for these controls; it is not a certification on its own."}], "meta": {"task": "chat", "os": "macos", "status": "refs", "dataset_version": "0.5.0"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 14 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: AI assistant access restricted [aiact.os_ai_features] in EU AI ACT CONTROLS\nSTATUS: Non-Compliant\nFOUND: assistant screen access on\nREQUIRED: screen access off unless approved\nDETAIL: Screen content can be sent to an external AI service.\nEVIDENCE: self-reported by the user\nREFERENCES:\n- ISO 27001:2022 A.5.34 Privacy and protection of PII: Meet the legal and contractual requirements for protecting personal data.\n- GDPR Art. 5(1)(c) Data minimisation: Personal data must be adequate, relevant and limited to what is necessary for the purpose.\n- GDPR Art. 25 Data protection by design and by default: Build data protection into systems and use privacy-friendly settings by default.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: Built-in OS AI features can capture screen content or send text to external AI services.\nrisk: Medium - Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\nfix (android):\n- In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\nverify: Re-run the audit; the OS AI features check should be Compliant or NotApplicable."}, {"role": "assistant", "content": "{\"check_id\": \"aiact.os_ai_features\", \"status\": \"Non-Compliant\", \"title\": \"AI assistant access restricted: does not meet the baseline\", \"finding\": \"AI assistant access restricted is assistant screen access on instead of screen access off unless approved. Built-in OS AI features can capture screen content or send text to external AI services. Self-reported by the user, not verified on the device.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Screen snapshots or prompts can contain personal and confidential data processed without a defined purpose.\"}, \"recommendation\": [\"In the Gemini app (or other assistant) settings, turn off screen context / access to app content unless the organisation has approved it.\"], \"references\": [\"ISO 27001:2022 A.5.34\", \"GDPR Art. 5(1)(c)\", \"GDPR Art. 25\", \"EU AI Act Art. 26\"], \"verification\": \"Re-run the audit; the OS AI features check should be Compliant or NotApplicable.\", \"evidence\": \"self-reported\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7001125, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "aiact.os_ai_features", "status": "Non-Compliant"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Android 15 (android)\nAUDIT RUN AS ADMIN: no\nCHECK: Human oversight for high-risk AI [org.human_oversight] in ORGANISATIONAL CONTROLS\nSTATUS: Pending\nFOUND: not answered\nREQUIRED: yes\nDETAIL: Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\nREFERENCES:\n- EU AI Act Art. 14 Human oversight: High-risk AI systems must be designed so that people can oversee them effectively.\n- EU AI Act Art. 26 Obligations of deployers of high-risk AI systems: Deployers must use high-risk AI as instructed, assign human oversight, monitor it, keep logs and inform affected people.\nGUIDANCE:\nwhy: High-risk AI systems need trained people who can understand, monitor and override them.\nrisk: Medium - High-risk AI without oversight can harm people without anyone noticing.\nfix (android):\n- For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\nverify: Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\nstatus note: This item needs a further step before it can be judged."}, {"role": "assistant", "content": "{\"check_id\": \"org.human_oversight\", \"status\": \"Pending\", \"title\": \"Human oversight for high-risk AI: needs a follow-up step\", \"finding\": \"Found: not answered. Answer ORG_HUMAN_OVERSIGHT in organisational.conf.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"Until resolved: High-risk AI without oversight can harm people without anyone noticing.\"}, \"recommendation\": [\"For each high-risk AI system, name a trained person responsible for oversight; answer n/a if no high-risk AI is used.\"], \"references\": [\"EU AI Act Art. 14\", \"EU AI Act Art. 26\"], \"verification\": \"Answer ORG_HUMAN_OVERSIGHT=yes (or n/a) in organisational.conf.\"}"}], "meta": {"task": "finding", "os": "android", "audit": 7000668, "profile": "neglected", "dataset_version": "0.5.0", "check_id": "org.human_oversight", "status": "Pending"}}
{"messages": [{"role": "system", "content": "You are the Devseis endpoint auditor. You write audit findings from collected evidence for ISO 27001, GDPR and the EU AI Act. Use only the evidence and reference material in the request. Never invent values, dates, versions, account names or tools. Reply with JSON only."}, {"role": "user", "content": "TASK: finding\nDEVICE: Ubuntu 20.04.6 LTS 20.04 (linux)\nAUDIT RUN AS ADMIN: yes\nCHECK: Removable media encryption [gdpr.removable_encryption] in GDPR TECHNICAL CONTROLS\nSTATUS: Non-Compliant\nFOUND: USB storage writable without enforced encryption\nREQUIRED: no unencrypted writes\nREFERENCES:\n- ISO 27001:2022 A.7.10 Storage media: Manage removable media through its life cycle according to the classification of the data on it.\n- ISO 27001:2022 A.8.24 Use of cryptography: Define and apply rules for using cryptography, including key management.\n- GDPR Art. 32(1)(a) Security of processing: pseudonymisation and encryption: Where appropriate, protect personal data with pseudonymisation and encryption.\n- GDPR Art. 34 Communication of a personal data breach to the data subject: Tell affected people about a high-risk breach without undue delay; encryption can remove this duty.\nGUIDANCE:\nwhy: Personal data copied to removable media must be encrypted, or copying must be blocked.\nrisk: Medium - A lost unencrypted USB drive with personal data is a reportable breach.\nfix (linux):\n- Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\nverify: Re-run the audit; Removable media encryption should be Compliant."}, {"role": "assistant", "content": "{\"check_id\": \"gdpr.removable_encryption\", \"status\": \"Non-Compliant\", \"title\": \"Removable media encryption: does not meet the baseline\", \"finding\": \"Found: USB storage writable without enforced encryption. Required: no unencrypted writes. Personal data copied to removable media must be encrypted, or copying must be blocked.\", \"risk\": {\"level\": \"Medium\", \"reason\": \"A lost unencrypted USB drive with personal data is a reportable breach.\"}, \"recommendation\": [\"Block USB storage writes (udev read-only rule or USBGuard) and provide approved encrypted drives.\"], \"references\": [\"ISO 27001:2022 A.7.10\", \"ISO 27001:2022 A.8.24\", \"GDPR Art. 32(1)(a)\", \"GDPR Art. 34\"], \"verification\": \"Re-run the audit; Removable media encryption should be Compliant.\"}"}], "meta": {"task": "finding", "os": "linux", "audit": 7000254, "profile": "hardened", "dataset_version": "0.5.0", "check_id": "gdpr.removable_encryption", "status": "Non-Compliant"}}